Neuroimaging Part II Joseph C.

Masdeu
And R. Gilberto González (Eds.)
Visit to download the full and correct content document:
https://ebookmass.com/product/neuroimaging-part-ii-joseph-c-masdeu-and-r-gilberto-
gonzalez-eds/
 HANDBOOK OF CLINICAL
NEUROLOGY

Series Editors

MICHAEL J. AMINOFF, FRANÇOIS BOLLER, AND DICK F. SWAAB

VOLUME 136

AMSTERDAM BOSTON HEIDELBERG LONDON NEW YORK OXFORD

PARIS SAN DIEGO SAN FRANCISCO SINGAPORE SYDNEY TOKYO
ELSEVIER
Radarweg 29, PO Box 211, 1000 AE Amsterdam, Netherlands
The Boulevard, Langford Lane, Kidlington, Oxford OX5 1GB, UK
50 Hampshire Street, 5th Floor, Cambridge, MA 02139, USA

© 2016 Elsevier B.V. All rights reserved.

No part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical,
including photocopying, recording, or any information storage and retrieval system, without permission in writing
from the publisher. Details on how to seek permission, further information about the Publisher’s permissions
policies and our arrangements with organizations such as the Copyright Clearance Center and the Copyright
Licensing Agency, can be found at our website: www.elsevier.com/permissions.

This book and the individual contributions contained in it are protected under copyright by the Publisher (other than
as may be noted herein).

Notices
Knowledge and best practice in this field are constantly changing. As new research and experience
broaden our understanding, changes in research methods, professional practices, or medical treatment may
become necessary.
Practitioners and researchers must always rely on their own experience and knowledge in evaluating and using any
information, methods, compounds, or experiments described herein. In using such information or methods they
should be mindful of their own safety and the safety of others, including parties for whom they have a professional
responsibility.
With respect to any drug or pharmaceutical products identified, readers are advised to check the most current
information provided (i) on procedures featured or (ii) by the manufacturer of each product to be administered, to
verify the recommended dose or formula, the method and duration of administration, and contraindications. It is
the responsibility of practitioners, relying on their own experience and knowledge of their patients, to make
diagnoses, to determine dosages and the best treatment for each individual patient, and to take all appropriate safety
precautions.
To the fullest extent of the law, neither the Publisher nor the authors, contributors, or editors, assume any liability
for any injury and/or damage to persons or property as a matter of products liability, negligence or otherwise, or
from any use or operation of any methods, products, instructions, or ideas contained in the material herein.

British Library Cataloguing-in-Publication Data

A catalogue record for this book is available from the British Library

Library of Congress Cataloging-in-Publication Data

A catalog record for this book is available from the Library of Congress

ISBN: 978-0-444-53486-6

For information on all Elsevier publications

visit our website at https://www.elsevier.com/

Publisher: Shirley Decker-lucke

Acquisition Editor: Mara Conner
Editorial Project Manager: Kristi Anderson
Production Project Manager: Sujatha Thirugnana Sambandam
Designer: Alan Studholme

Typeset by SPi Global, India

Handbook of Clinical Neurology 3rd Series

Available titles
Vol. 79, The human hypothalamus: basic and clinical aspects, Part I, D.F. Swaab, ed. ISBN 9780444513571
Vol. 80, The human hypothalamus: basic and clinical aspects, Part II, D.F. Swaab, ed. ISBN 9780444514905
Vol. 81, Pain, F. Cervero and T.S. Jensen, eds. ISBN 9780444519016
Vol. 82, Motor neurone disorders and related diseases, A.A. Eisen and P.J. Shaw, eds. ISBN 9780444518941
Vol. 83, Parkinson’s disease and related disorders, Part I, W.C. Koller and E. Melamed, eds. ISBN 9780444519009
Vol. 84, Parkinson’s disease and related disorders, Part II, W.C. Koller and E. Melamed, eds. ISBN 9780444528933
Vol. 85, HIV/AIDS and the nervous system, P. Portegies and J. Berger, eds. ISBN 9780444520104
Vol. 86, Myopathies, F.L. Mastaglia and D. Hilton Jones, eds. ISBN 9780444518996
Vol. 87, Malformations of the nervous system, H.B. Sarnat and P. Curatolo, eds. ISBN 9780444518965
Vol. 88, Neuropsychology and behavioural neurology, G. Goldenberg and B.C. Miller, eds. ISBN 9780444518972
Vol. 89, Dementias, C. Duyckaerts and I. Litvan, eds. ISBN 9780444518989
Vol. 90, Disorders of consciousness, G.B. Young and E.F.M. Wijdicks, eds. ISBN 9780444518958
Vol. 91, Neuromuscular junction disorders, A.G. Engel, ed. ISBN 9780444520081
Vol. 92, Stroke – Part I: Basic and epidemiological aspects, M. Fisher, ed. ISBN 9780444520036
Vol. 93, Stroke – Part II: Clinical manifestations and pathogenesis, M. Fisher, ed. ISBN 9780444520043
Vol. 94, Stroke – Part III: Investigations and management, M. Fisher, ed. ISBN 9780444520050
Vol. 95, History of neurology, S. Finger, F. Boller and K.L. Tyler, eds. ISBN 9780444520081
Vol. 96, Bacterial infections of the central nervous system, K.L. Roos and A.R. Tunkel, eds. ISBN 9780444520159
Vol. 97, Headache, G. Nappi and M.A. Moskowitz, eds. ISBN 9780444521392
Vol. 98, Sleep disorders Part I, P. Montagna and S. Chokroverty, eds. ISBN 9780444520067
Vol. 99, Sleep disorders Part II, P. Montagna and S. Chokroverty, eds. ISBN 9780444520074
Vol. 100, Hyperkinetic movement disorders, W.J. Weiner and E. Tolosa, eds. ISBN 9780444520142
Vol. 101, Muscular dystrophies, A. Amato and R.C. Griggs, eds. ISBN 9780080450315
Vol. 102, Neuro-ophthalmology, C. Kennard and R.J. Leigh, eds. ISBN 9780444529039
Vol. 103, Ataxic disorders, S.H. Subramony and A. Durr, eds. ISBN 9780444518927
Vol. 104, Neuro-oncology Part I, W. Grisold and R. Sofietti, eds. ISBN 9780444521385
Vol. 105, Neuro-oncology Part II, W. Grisold and R. Sofietti, eds. ISBN 9780444535023
Vol. 106, Neurobiology of psychiatric disorders, T. Schlaepfer and C.B. Nemeroff, eds. ISBN 9780444520029
Vol. 107, Epilepsy Part I, H. Stefan and W.H. Theodore, eds. ISBN 9780444528988
Vol. 108, Epilepsy Part II, H. Stefan and W.H. Theodore, eds. ISBN 9780444528995
Vol. 109, Spinal cord injury, J. Verhaagen and J.W. McDonald III, eds. ISBN 9780444521378
Vol. 110, Neurological rehabilitation, M. Barnes and D.C. Good, eds. ISBN 9780444529015
Vol. 111, Pediatric neurology Part I, O. Dulac, M. Lassonde and H.B. Sarnat, eds. ISBN 9780444528919
Vol. 112, Pediatric neurology Part II, O. Dulac, M. Lassonde and H.B. Sarnat, eds. ISBN 9780444529107
Vol. 113, Pediatric neurology Part III, O. Dulac, M. Lassonde and H.B. Sarnat, eds. ISBN 9780444595652
Vol. 114, Neuroparasitology and tropical neurology, H.H. Garcia, H.B. Tanowitz and O.H. Del Brutto, eds.
ISBN 9780444534903
Vol. 115, Peripheral nerve disorders, G. Said and C. Krarup, eds. ISBN 9780444529022
Vol. 116, Brain stimulation, A.M. Lozano and M. Hallett, eds. ISBN 9780444534972
Vol. 117, Autonomic nervous system, R.M. Buijs and D.F. Swaab, eds. ISBN 9780444534910
Vol. 118, Ethical and legal issues in neurology, J.L. Bernat and H.R. Beresford, eds. ISBN 9780444535016
Vol. 119, Neurologic aspects of systemic disease Part I, J. Biller and J.M. Ferro, eds. ISBN 9780702040863
Vol. 120, Neurologic aspects of systemic disease Part II, J. Biller and J.M. Ferro, eds. ISBN 9780702040870
Vol. 121, Neurologic aspects of systemic disease Part III, J. Biller and J.M. Ferro, eds. ISBN 9780702040887
Vol. 122, Multiple sclerosis and related disorders, D.S. Goodin, ed. ISBN 9780444520012
Vol. 123, Neurovirology, A.C. Tselis and J. Booss, eds. ISBN 9780444534880
vi AVAILABLE TITLES (Continued)
Vol. 124, Clinical neuroendocrinology, E. Fliers, M. Korbonits and J.A. Romijn, eds. ISBN 9780444596024
Vol. 125, Alcohol and the nervous system, E.V. Sullivan and A. Pfefferbaum, eds. ISBN 9780444626196
Vol. 126, Diabetes and the nervous system, D.W. Zochodne and R.A. Malik, eds. ISBN 9780444534804
Vol. 127, Traumatic brain injury Part I, J.H. Grafman and A.M. Salazar, eds. ISBN 9780444528926
Vol. 128, Traumatic brain injury Part II, J.H. Grafman and A.M. Salazar, eds. ISBN 9780444635211
Vol. 129, The human auditory system: Fundamental organization and clinical disorders, G.G. Celesia
and G. Hickok, eds. ISBN 9780444626301
Vol. 130, Neurology of sexual and bladder disorders, D.B. Vodušek and F. Boller, eds. ISBN 9780444632470
Vol. 131, Occupational neurology, M. Lotti and M.L. Bleecker, eds. ISBN 9780444626271
Vol. 132, Neurocutaneous syndromes, M.P. Islam and E.S. Roach, eds. ISBN 9780444627025
Vol. 133, Autoimmune neurology, S.J. Pittock and A. Vincent, eds. ISBN 9780444634320
Vol. 134, Gliomas, M.S. Berger and M. Weller, eds. ISBN 9780128029978
Vol. 135, Neuroimaging Part I, J.C. Masdeu and R.G. González, eds. ISBN 9780444534859
 Foreword

We are proud to present the first volumes dedicated to neuroimaging in the Handbook of Clinical Neurology series.
Neurologists, not just those in training, may wonder at the kind of medical world that existed before modern imaging.
Indeed, the neuroscience community has since its beginning attempted to understand the human mind and brain
through imaging. As far back as 1880, the Italian physiologist Angelo Mosso introduced the “human circulation
balance,” which could noninvasively measure the redistribution of blood during emotional and intellectual activity.
More recently, semi-invasive techniques such as pneumoencephalography (introduced by Dandy in 1918) and arteri-
ography (pioneered by Moniz in 1927) allowed partial visualization of the brain and its surrounding structures. New
methods enabling easier, safer, noninvasive, painless, and repeatable imaging have only been developed in the past
50 years or so, starting with computed tomography, some of whose developers won the 1979 Nobel Prize for medicine
or physiology. The many subsequent developments in neuroimaging are masterfully presented in these two volumes.
The volumes deal with a variety of neuroimaging-related topics. They include thorough descriptions of the involved
methods and their application to specific diseases of the brain, spinal cord, and peripheral nervous system. Emphasis is
given to the most common disorders, such as tumors, strokes, multiple sclerosis, movement disorders, infections,
dementia, and trauma, but less common conditions such as neurocutaneous syndromes are also discussed. The impor-
tant questions of when and where to image, as well as the differential diagnosis of imaging findings, are discussed in
the light of specific syndromes. A separate section covers pediatric neuroimaging. The volumes conclude with sections
dedicated to interventional neuroimaging as well as to postmortem imaging and neuropathologic correlations.
We have been fortunate to have as volume editors two distinguished scholars, Dr. Joseph C. Masdeu, of the Depart-
ment of Neurology, Methodist Hospital, Houston, Texas, and Dr. R. Gilberto González, from the Department of Radi-
ology, Massachusetts General Hospital in Boston. Both have been at the forefront of neuroimaging research for many
years. They have assembled a truly international group of authors with acknowledged expertise to contribute to the
texts and have produced two authoritative, comprehensive, and up-to-date volumes. Their availability electronically on
Elsevier’s Science Direct site as well as in print format should ensure their ready accessibility and facilitate searches for
specific information.
We are grateful to the volume editors and to all the contributors for their efforts in creating such an invaluable
resource. As series editors we read and commented on each of the chapters with great interest. We are therefore con-
fident that both clinicians and researchers in many different medical disciplines will find much in these volumes to
appeal to them.
And last, but not least, it is always a pleasure to acknowledge and thank Elsevier, our publisher – and, in particular,
Michael Parkinson in Scotland, and Mara Conner and Kristi Anderson in San Diego – for their unfailing and expert
assistance in the development and production of these volumes.
Michael J. Aminoff
François Boller
Dick F. Swaab
 Preface

Neuroimaging has become one of the most useful set of tools for understanding and diagnosing diseases of the ner-
vous system. Fittingly, the present two volumes of the Handbook of Clinical Neurology review the extensive advances
in the field. In the first volume, discussions of the various techniques used in neuroimaging are followed by reviews of
the imaging findings caused by brain diseases. We have chosen not to include a chapter on brain anatomy because it
would be quite long and extant atlases are excellent. The second volume begins with a description of the functional
anatomy of the spine and of the imaging findings in diseases of the spine and spinal cord. Imaging of peripheral nerve
and muscle follows. Then, there is a section on when and how to image the various clinical syndromes produced by
diseases of the nervous system. Adequacy in the use of expensive neuroimaging tools has always been a priority, but it
is becoming more acute as the application of neuroimaging expands more rapidly than the available resources. The
next section is unusual in a book of this type: a description of the various imaging findings that should lead to con-
sideration of the diseases causing them. Such information is particularly important when using techniques like com-
puted tomography and magnetic resonance imaging, which offer a panoply of findings and are extensively used in
clinical practice. Next is a section on pediatric neuroimaging, led by a chapter on imaging findings during normal
development. After three chapters on the therapeutic use of endovascular imaging, the second volume concludes with
a chapter on postmortem imaging as a tool to better define normal brain structure on imaging and its alteration by
some disorders.
We hope that this book will be useful to all those who work with clinical imaging of the nervous system, such as
neurologists, neuroradiologists, neurosurgeons, and nuclear medicine physicians. Some sections, for instance, the sec-
tions on the spine, peripheral nerve, and muscle, may be helpful to orthopedic surgeons and rehabilitation specialists.
Neuropsychologists may find helpful the chapters on neurodegenerative disorders leading to cognitive impairment.
Neuroimaging is used not only clinically, but also by those interested in clarifying the still largely undiscovered
landscape and functional intricacy of the brain. While the clinical literature of neuroimaging is extensive, even more
extensive and more widely cited is the literature of neuroimaging applied to the study of the healthy human nervous
system. Although human disease has traditionally led to a better understanding of normal structure and function,
researchers looking primarily for information on the normal nervous system should look elsewhere.
We are most thankful to the authors, who have distilled their expertise in superbly written and illustrated chapters.
Mr. Michael Parkinson, from Elsevier, has skillfully coordinated the gathering of information for these two volumes.
We are also thankful to the three series editors and, particularly, to Dr. François Boller, for their excellent suggestions.
Joseph C. Masdeu
R. Gilberto González
 Contributors
F. Agosta
Neuroimaging Research Unit, Institute of Experimental
Neurology, Division of Neuroscience, San Raffaele
Scientific Institute, Vita-Salute San Raffaele University,
Milan, Italy
C. Amlie-Lefond
Department of Neurology, Seattle Children’s Hospital,
Seattle, WA, USA
A. Atri
Ray Dolby Brain Health Center, California Pacific
Medical Center Research Institute, Sutter Health,
San Francisco, CA, USA
J.C. Augustinack
Athinoula A. Martinos Center for Biomedical Imaging,
Department of Radiology, Massachusetts General
Hospital, Charlestown, MA, USA
A. Bali
Department of Radiology, Antwerp University Hospital
and University of Antwerp, Antwerp, Belgium
A.M. Blitz
Neuro-radiology Division, Johns Hopkins University
School of Medicine, Baltimore, MD, USA
N. Bogduk
Newcastle Bone and Joint Institute, University of
Newcastle, Newcastle, Australia
A. Boon
Department of Physical Medicine and Rehabilitation and
Department of Neurology, Mayo Clinic, Rochester, MN,
USA
B.H. Brinkmann
Division of Epilepsy, Department of Neurology, Mayo
Clinic, Rochester, MN, USA
H. Brunel
Neuroradiology Service, H^opital la Timone, Marseille,
France
N.D. Bryant
Vanderbilt University Institute of Imaging Science and
the Department of Radiology and Radiological Sciences,
Vanderbilt University, Nashville, TN, USA
P.M. Capone
Medical Imaging and Neurology, Winchester Medical
Center, Winchester and Department of Neurology,
Virginia Commonwealth University, Richmond, VA, USA
G.D. Cascino
Division of Epilepsy, Department of Neurology, Mayo
Clinic, Rochester, MN, USA
M. Castillo
Division of Neuroradiology, Department of Radiology,
University of North Carolina, Chapel Hill, NC, USA
F. Cendes
University of Campinas, Department of Neurology,
Campinas, SP, Brazil
C.T. Chin
Department of Radiology, University of California,
San Francisco, CA, USA
H.M. Dahmoush
Department of Radiology, Children’s Hospital of
Philadelphia and University of Pennsylvania,
Philadelphia, PA, USA
B.M. Damon
Vanderbilt University Institute of Imaging Science and
the Department of Radiology and Radiological Sciences,
Departments of Biomedical Engineering and Molecular
Physiology and Biophysics, Vanderbilt University,
Nashville, TN, USA
xii CONTRIBUTORS
C.C. Della Santina P. Hedera
Department of Otolaryngology, Head and Neck Department of Neurology, Vanderbilt University,
Surgery, Johns Hopkins University School of Medicine, Nashville, TN, USA
Baltimore, MD, USA
J.A. Hirsch
O. Diaz Neuroendovascular Service, Massachusetts General
Neurovascular Center, Methodist Hospital, Houston, Hospital, Boston, MA, USA
TX, USA
T.A.G.M. Huisman
C. Ellenberger Division of Pediatric Radiology, Johns Hopkins
Lebanon Magnetic Imaging, Lebanon, Pennsylvania, Hospital, Baltimore, MD, USA
USA
J.V. Hunter
Department of Pediatric Radiology, Texas Children’s
A. Faje
Hospital and Baylor College of Medicine, Houston, TX,
Neuroendocrine Unit, Massachusetts General Hospital
USA
and Harvard Medical School, Boston, MA, USA
M. Kellogg
A. Faridar Department of Neurology, Oregon Health and Science
Department of Neurology, Houston Methodist Hospital, University, Portland, OR, USA
Houston, TX, USA
F.J. Kirkham
M. Filippi Neurosciences Unit, UCL Institute of Child Health,
Neuroimaging Research Unit, Institute of Experimental London; Southampton Children’s Hospital and Clinical
Neurology, Division of Neuroscience, San Raffaele and Experimental Sciences, University of Southampton,
Scientific Institute, Vita-Salute San Raffaele University, Southampton, UK
Milan, Italy
J.P. Klein
C.R. Fitz Departments of Neurology and Radiology, Brigham and
Department of Radiology, Children’s Hospital of Women’s Hospital and Harvard Medical School,
Pittsburgh, Pittsburgh, PA, USA Boston, MA, USA

A.E. Flanders A. Klibanski

Department of Radiology, Thomas Jefferson University Neuroendocrine Unit, Massachusetts General Hospital
Hospital, Philadelphia, PA, USA and Harvard Medical School, Boston, MA, USA

A.D. Furtado M. Kliot

Department of Radiology, Children’s Hospital of Department of Neurological Surgery, Northwestern
Pittsburgh, Pittsburgh, PA, USA Feinberg School of Medicine, Chicago, IL, USA

R. Gadhia M. Koob
Department of Neurology, Houston Methodist Hospital, Pediatric Radiology Imaging Service, Centre Hospitalier
Houston, TX, USA Universitaire de Strasbourg, H^opital de Hautepierre and
Laboratoire ICube, Universite de Strasbourg-CNRS,
N. Girard Strasbourg, France
Neuroradiology Service, H^ opital la Timone and Aix
Marseille Universite, Marseille, France T. Krings
Division of Neuroradiology and Neurosurgery,
L. Goerner University of Toronto, Toronto Western Hospital and
The Radiology Group, Honolulu, HI, USA University Health Network, Toronto, ON, Canada

R.G. González T. Leslie-Mazwi

Department of Radiology, Massachusetts General Neuroendovascular Service, Massachusetts General
Hospital, Boston, MA, USA Hospital, Boston, MA, USA
 CONTRIBUTORS xiii
K. Li P.M. Parizel
Vanderbilt University Institute of Imaging Science and Department of Radiology, Antwerp University Hospital
the Department of Radiology and Radiological Sciences, and University of Antwerp, Antwerp, Belgium
Vanderbilt University, Nashville, TN, USA
T.J. Pfiffner
C.W. Liang Dent Neurologic Institute, Amherst, NY, USA
Neurovascular Imaging Research Core and Department
of Neurology, University of California Los Angeles, S. Pillen
Los Angeles, CA, USA Department of Sleep Medicine, Kempenhaeghe
Expertise Center for Epileptology, Sleep Medicine and
D.S. Liebeskind Neurocognition, Heeze, The Netherlands
Neurovascular Imaging Research Core and Department
N.K. Pinter
of Neurology, University of California Los Angeles,
Dent Neurologic Institute, Amherst, NY, USA
Los Angeles, CA, USA
A. Poretti
B.O. Marquis
Division of Pediatric Radiology, Johns Hopkins
Division of Epilepsy, Department of Neurology, State
Hospital, Baltimore, MD, USA
University of New York Downstate Medical Center,
Brooklyn, NY, USA
J. Rabinov
Neuroendovascular Service, Massachusetts General
J.C. Masdeu
Hospital, Boston, MA, USA
Department of Neurology, Houston Methodist Hospital,
Houston, TX, USA
L. Rangel-Castilla
Neuroendovascular Surgery, University of Buffalo
M.J. McKeown
Neurosurgery, State University of New York, Buffalo,
Pacific Parkinson’s Research Centre and Division of
NY, USA
Neurology, Djavad Mowafaghian Centre for Brain
Health, University of British Columbia and Vancouver
H.L. Rekate
Coastal Health, Vancouver, BC, Canada
Department of Neurosurgery, Hofstra Northshore LIJ
School of Medicine, Manhasset, and The Chiari
L.L. Mechtler
Institute, Great Neck, NY, USA
Dent Neurologic Institute, Amherst, NY, USA
D. Rodriguez-Luna
E.R. Melhem
Stroke Unit, Department of Neurology, Vall d’Hebron
Department of Diagnostic Radiology and Nuclear
University Hospital, Vall d’Hebron Research Institute,
Medicine, University of Maryland, Baltimore, MD, USA
Autonomous University of Barcelona, Barcelona, Spain
C.A. Molina
R. Ruff
Stroke Unit, Department of Neurology, Vall d’Hebron
Department of Neurology, Case Western Reserve
University Hospital, Vall d’Hebron Research Institute,
University School of Medicine, Cleveland, OH, USA
Autonomous University of Barcelona, Barcelona, Spain
G. Saliou
D.E. Newman-Toker Neuroradiology Service, Centre Hospitalier
Department of Neurology and Department of Universitaire Bic^etre, Le Kremlin Bic^etre, France
Otolaryngology, Head and Neck Surgery, Johns Hopkins
University School of Medicine, Baltimore, MD, USA R. Scranton
Department of Neurosurgery, Methodist Hospital,
R.B. Olatunji Houston, TX, USA
Department of Radiology, University Hospital Vienna,
Vienna, Austria S.M. Selkirk
Spinal Cord Injury Division, Louis Stokes Cleveland
A. Panigrahy Veterans Affairs Medical Center and Department of
Department of Radiology, Children’s Hospital of Neurology, Case Western Reserve University School of
Pittsburgh, Pittsburgh, PA, USA Medicine, Cleveland, OH, USA
xiv CONTRIBUTORS
P. Shankar
Division of Neuroradiology, Department of Radiology,
University of North Carolina, Chapel Hill, NC, USA
D. Shaw
Department of Radiology, Seattle Children’s Hospital,
Seattle, WA, USA
N.G. Simon
St Vincent’s Clinical School, University of New South
Wales, Sydney, Australia
A.J. Stoessl
Pacific Parkinson’s Research Centre and Division of
Neurology, Djavad Mowafaghian Centre for Brain
Health, University of British Columbia and Vancouver
Coastal Health, Vancouver, BC, Canada
N.A. Tritos
Neuroendocrine Unit, Massachusetts General
Hospital and Harvard Medical School, Boston, MA,
USA
N. van Alfen
Department of Neurology, Radboud University Medical
Center, Nijmegen, The Netherlands
L. van den Hauwe
Department of Radiology, Antwerp University Hospital
and University of Antwerp, Antwerp, Belgium
A.J.W. van der Kouwe
Athinoula A. Martinos Center for Biomedical Imaging,
Department of Radiology, Massachusetts General
Hospital, Charlestown, MA, USA

V. Sulc J. Van Goethem

Department of Neurology, 2nd Faculty of Medicine,
Charles University in Prague and Motol University
Hospital, Czech Republic
B. Swearingen
Department of Neurosurgery, Massachusetts General
Hospital and Harvard Medical School, Boston, MA, USA
J. Talbott
Department of Radiology, University of California,
San Francisco, CA, USA
Department of Radiology, Antwerp University Hospital
and University of Antwerp, Antwerp, Belgium
A.J.L. Van Hoyweghen
Department of Radiology, Antwerp University Hospital
and University of Antwerp, Antwerp, Belgium
A. Vossough
Department of Radiology, Children’s Hospital of
Philadelphia and University of Pennsylvania,
Philadelphia, PA, USA

W.H. Theodore C. Zamora

National Institute of Neurological Disorders and Stroke, Division of Neuroradiology, Department of Radiology,
Bethesda, MD, USA University of North Carolina, Chapel Hill, NC, USA

M.M. Thurnher V.M. Zohrabian

Department of Radiology, University Hospital Vienna, Department of Diagnostic Radiology, Yale University
Vienna, Austria School of Medicine, New Haven, CT, USA
Handbook of Clinical Neurology, Vol. 136 (3rd series)
Neuroimaging, Part II
J.C. Masdeu and R.G. González, Editors
© 2016 Elsevier B.V. All rights reserved

Chapter 32

Functional anatomy of the spine

NIKOLAI BOGDUK*
Newcastle Bone and Joint Institute, University of Newcastle, Newcastle, Australia

Abstract
Among other important features of the functional anatomy of the spine, described in this chapter, is the
remarkable difference between the design and function of the cervical spine and that of the lumbar spine.
In the cervical spine, the atlas serves to transmit the load of the head to the typical cervical vertebrae. The
axis adapts the suboccipital region to the typical cervical spine. In cervical intervertebrtal discs the anulus
fibrosus is not circumferential but is crescentic, and serves as an interosseous ligament in the saddle joint
between vertebral bodies. Cervical vertebrae rotate and translate in the sagittal plane, and rotate in the
manner of an inverted cone, across an oblique coronal plane. The cervical zygapophysial joints are the
most common source of chronic neck pain. By contrast, lumbar discs are well designed to sustain com-
pression loads, but rely on posterior elements to limit axial rotation. Internal disc disruption is the most
common basis for chronic low-back pain. Spinal muscles are arranged systematically in prevertebral and
postvertebral groups. The intrinsic elements of the spine are innervated by the dorsal rami of the spinal
nerves, and by the sinuvertebral nerves. Little modern research has been conducted into the structure of
the thoracic spine, or the causes of thoracic spinal pain.

INTRODUCTION CERVICAL SPINE

In writing a chapter on anatomy for neurologists the risk
arises of being arcane or banal. Neurologists will already
be familiar with the precepts of classic anatomy, and
would not be inclined to read a chapter that repeats bor-
ing, undergraduate material. For these reasons, the pre-
sent chapter has been cast in a different manner.
Although conventional elements of anatomy are
reprised, they are permeated by several themes. New
facts are provided, stemming from modern research into
the structure of the spine, along with new perceptions
about design and function. Throughout, the focus is
on clinical relevance, particularly with respect to the
mechanisms of spinal injury and spinal pain. In that
regard, certain structures – ignored in conventional
undergraduate curricula – are promoted to epidemiolog-
ically significant, clinical importance.
The cervical spine serves as a mobile support for the sen-
sory platform of the head. It allows the sensory apparatus
for vision, hearing, and smell to be elevated or depressed
in the sagittal plane, and to scan the environment in the
horizontal plane. In order to subserve these functions,
the cervical spine has to be mobile, yet sufficiently strong
to support the weight of the head. Its vulnerability, to
either minor or major injuries, lies in being long, slender,
and carrying the large mass of the head at its summit.
Both for descriptive purposes and functionally, the
cervical spine can be divided into three zones: the suboc-
cipital zone, centered on the C1 vertebra; a transitional
zone formed by the C2 vertebra; and the typical zone,
encompassing the C–7 vertebrae (Bogduk and Mercer,
2000) (Fig. 32.1). These zones differ both in structure
and in function.

*Correspondence to: Nikolai Bogduk, PO Box 128, The Junction, New South Wales 2291, Australia. E-mail: nbogduk@bigpond.
net.au
676 N. BOGDUK

Fig. 32.1. Sagittal magnetic resonance images of the cervical spine, showing its structure and zones. (A) Median scan, showing
the vertebral bodies and interverterbral discs. The white dots mark the mean location of the axes of rotation for flexion-extension of
the vertebra above. The odontoid process (op) projects rostrally from the body of C2, to lie behind the anterior arch of the atlas (C1).
(B) Lateral scan, showing the occipital condyle (oc), the lateral mass (lm) of the atlas (C1), the articular pillars (ap), and the zyga-
pophysial joints (zj) that they form, at the segments labeled. (Courtesy of Dr. Tim Maus, Mayo Clinic, Rochester, MN.)

Suboccipital zone Upper transition zone

The C1 vertebra (the atlas) shares none of the features of
typical cervical vertebrae, and should never have been
considered cervical. In structure and in function it is
more like an occipital vertebra. In structure it resembles
the occipital bone, as can be seen in axial scans. In func-
tion, it more closely operates with the head, rather than
with the remainder of the cervical spine.
The classic description of the atlas as a ring vertebra
belies its design and function. The critical components of
the atlas are its two lateral masses (Fig. 32.2). Superiorly,
these present superior articular processes that receive the
occipital condyles, and thereby cradle the skull. Inferi-
orly, the lateral masses present inferior articular pro-
cesses that rest on the C2 vertebra, and thereby
transmit the load of the head to the remainder of the cer-
vical spine. The anterior and posterior arches of the atlas
serve little function other than holding the two lateral
masses both apart and together, while the latter do the
mechanical work of the atlas.
Upon receiving the occipital condyles into their deep
sockets, the superior articular processes of each lateral
mass form the atlanto-occipital joints (Figs 32.2 and
32.3). These synovial joints constitute the only direct con-
nection between the skull and C1. They allow a small
range of flexion-extension, but the depth of their sockets
precludes axial rotation. Therefore, as the head rotates
(in the transverse plane) the atlas is obliged to move with
it. In that respect, the atlas behaves like a passive washer,
between the skull and C2.
The upper half of the C2 vertebra (the axis) is designed to
support the atlas. Superiorly and laterally, it presents
superior articular processes that slope caudally and lat-
erally, and act like sloping shoulders on which the lateral
masses of the atlas rest (Fig. 32.2). The inferior articular
processes of the atlas have a reciprocal, caudal and lat-
eral slope. The apposed articular processes on each side
form the lateral atlantoaxial joint (Figs 32.2 and 32.3).
The caudolateral slope of the lateral atlantoaxial joint
helps stabilize the atlas in the coronal plane, but also
underlies the mechanism of Jefferson fractures. Severe
axial loads, applied to the skull, will drive the atlas cau-
dally, but its lateral masses will also spread laterally
down the lateral slope of the lateral atlantoaxial joints,
resulting in burst fractures of the anterior and posterior
arches.
Centrally, the axis presents a long odontoid process
(the dens) that projects behind the anterior arch of the
atlas, with which it forms the median atlantoaxial joint
(Figs 32.1, 32.3, and 32.4). The anterior arch is held against
the odontoid process by the transverse ligament, which
spans like a belt between the two lateral masses of the
atlas, behind the odontoid process (Figs 32.4 and 32.5).
Posterior displacement of the atlas is prevented by
impaction of the anterior arch against the odontoid pro-
cess, at the median atlantoaxial joint. Anterior displace-
ment is prevented by tension in the transverse ligament
(Fielding et al., 1974). The ligament allows up to 3 mm
normal range of separation between the odontoid
 FUNCTIONAL ANATOMY OF THE SPINE 677

Fig. 32.2. Coronal magnetic resonance images of the cervical spine, showing the structure of its components. (A) Anterior scan,
showing the occipital condyles (oc) resting in the sockets of the lateral masses (lm) of the atlas, and forming the atlanto-occipital
joints (aoj); and the lateral masses bracketing the odontoid process (op), and resting on the “shoulders” of the axis (C2), where they
form the lateral atlantoaxial joints (laaj). The vertebral bodies of C2–7 form the anterior column of the cervical spine. (B) Posterior
scan, through the synovial joints of the cervical spine. Note the wedge shape of the lateral mass (lm) between the atlanto-occipital
joint (aoj) and the lateral atlantoaxial joint (laaj). The zygapophysial joint at C2–3 slopes caudally and medially, but those at suc-
cessive levels are essentially horizontal. The dotted line illustrates the ellipsoid shape depicted by the C2–3 zygapophysial joints
and the C2–3 disc, into which the atlas (C2) nestles on to the typical cervical spine.

process and the anterior arch in adults, and 5 mm in chil-
dren. In the past, the magnitude of the interval between
the anterior arch and the odontoid process has been used
as a measure of atlantoaxial instability, but as a predictor
of neurologic compromise the posterior atlantodental
interval (Fig. 32.3) has greater sensitivity and specificity
(Wasserman et al., 2011).
Severe forces delivered anteriorly to the head can frac-
ture the odontoid process. Such fractures threaten the sag-
ittal stability of the atlas. In turn, anterior or posterior
displacement of the atlas can threaten the spinal cord.
Rheumatoid arthritis of the atlantoaxial joints can weaken
the transverse ligament of the atlas, resulting in anterior
subluxation of the atlas (Wasserman et al., 2011).

Fig. 32.3. Close-up views of sagittal magnetic resonance images of the suboccipital joints. (A) Median scan through the odon-
toid process (op) and vertebral body of C2. With the front of the odontoid process, the anterior arch (aa) of the atlas forms the
median atlantoaxial joint (maaj). The transverse arrow marks the posterior atlantodental (pa) interval. (B) Lateral scan through
the lateral mass (lm) of the atlas. With the superior sockets of the lateral mass of the atlas, the occipital condyle (oc) forms the
atlanto-occipital joint (aoj). With C2, the lateral mass forms the lateral atlantoaxial joint (laaj). Note the bi-convex shape of
the lateral atlantoaxial joint. The triangular, white signals anteriorly and posteriorly within the joint are the fibroadipose meniscoids
that it contains. (Courtesy of Dr. Tim Maus, Mayo Clinic, Rochester MN.)
678 N. BOGDUK
Fig. 32.4. A sketch of an axial (top) view of the atlas sitting on
the axis. The lateral masses of the atlas bracket the odontoid
process (op) of the axis, and are themselves joined by the ante-
rior (aa) and posterior (pa) arches of the atlas. With the anterior
arch, the odontoid process forms the median atlantoaxial joint
(maaj). The transverse ligament (tl) spans like a belt behind the
odontoid processes, between the two lateral masses.
Although the osseous articular processes of the lat-
eral atlantoaxial joint are flat, their articular cartilages
are convex (along the sagittal plane) (Koebke and
Brade, 1982) (Fig. 32.3). As a result, the atlas perches
somewhat precariously on the axis, with its convex infe-
rior articular cartilages balancing on the convexities of
the superior articular cartilages of the axis. The spaces
anteriorly and posteriorly between the convex cartilages
are filled by wedge-shaped fibrocartilaginous menis-
coids (Mercer and Bogduk, 1993).
Although a small degree of flexion and extension is
possible between the atlas and the axis (Bogduk and
Mercer, 2000), the cardinal movement between these
two vertebrae is axial rotation. During this movement
the atlas pivots at the median atlantoaxial joint, while
its lateral masses slide backwards or forwards, circum-
ferentially, at the lateral atlantoaxial joints. However,
because of the convexity of the articular cartilages in
these joints, the lateral masses also descend, down the
posterior or anterior slope of the cartilages, as they move
backwards or forwards, respectively. As a result, the
atlas settles (lowers or screws down) during axial rota-
tion, and rises when the movement is reversed
(Bogduk and Mercer, 2000). During these displace-
ments, the meniscoids of the joints cover the exposed
surfaces of the subluxating articular cartilages.
The total range of axial rotation of the atlas is consid-
erably large. It has been measured as 43  5.5° which
effectively amounts to 50% of the range of axial rotation
of the head and neck (Bogduk and Mercer, 2000). At the
extremes of this range, very little of the articular
Fig. 32.5. A sketch of the suboccipital joints and ligaments, as
view from behind with the posterior arch of the atlas resected.
The lateral mass (lm) of the atlas supports the occipital bone,
and rests of the axis, forming the atlanto-occipital joint (aoj)
above, and the lateral atlantoaxial joint (laaj) below. The trans-
verse ligament (tl) holds the atlas (al) against the odontoid pro-
cess (op). The alar ligaments bind the odontoid process to the
margins of the foramen magnum, thereby connecting the skull
to C2 but bypassing C1.
cartilages of the lateral atlantoaxial joints remain
opposed; the joint is almost dislocated.
In order to accommodate this large range of motion,
the capsules of the lateral atlantoaxial joint are loose,
and serve little to hold the atlas on the axis. That service
is provided by the alar ligaments. On each side, these lig-
aments pass essentially transversely from the upper end
of the odontoid process to the margin of the foramen
magnum (Fig. 32.5). In doing so, they bypass the atlas,
and lock the head into place on the axis, effectively clamp-
ing the atlas between the skull and C2. The alar ligaments
are the cardinal restraint to axial rotation of the head
(Dvorak et al., 1987). They are sufficiently strong to pre-
vent anterior dislocation of the head even if the transverse
ligament is completely severed (Fielding et al., 1974). Dis-
ruption of an alar ligament can result in rotatory instabil-
ity of the head and atlas (Dvorak et al., 1987).
Excessive axial rotation of the atlas can result in a lat-
eral mass dislocating at the lateral atlantoaxial joint,
causing fixed atlantoaxial deformity (Wortzman and
Dewar, 1968). A less dramatic form of torticollis can
arise if, after rotation of the head and atlas, a meniscoid
of the lateral atlantoaxial joint fails to re-enter the joint
space, catches under the capsule of the joint, and acts like
a loose body to prevent derotation of the joint (Mercer
and Bogduk, 1993).
 FUNCTIONAL ANATOMY OF THE SPINE
Fig. 32.6. A sketch of a coronal view of how forces from the
head are transmitted into the cervical spine. On each side, the
weight of the head passes through the occipital condyle (oc),
into the lateral mass (lm) of the atlas, and into the axis
(C2) through the lateral atlantoaxial joint. From there, the forces
diverge, partly into the posterior column of zygapophysial
joints, and partly into the anterior column of vertebral bodies
and discs. Half the load passes anteriorly and half posteriorly.
Increasing interest has been focused on the lateral
atlantoaxial joints as a possible source of cervicogenic
headache. This contention can be tested by controlled,
intra-articular, diagnostic blocks of the putatively pain-
ful joint (Bogduk and Bartsch, 2008; Bogduk and
Govind, 2009; Bogduk, 2014).
Lower transition zone
The lower half of the C2 vertebra has the structure of a
typical cervical vertebra (Figs 32.1 and 32.2). Centrally it
presents a vertebral body, and laterally it presents paired
inferior articular processes. Having received the lateral
masses of the atlas, the axis transmits the load of the
head along an anterior channel, through its vertebral
body to the vertebral bodies below, and along paired pos-
terior channels, through the zygapophysial joints
(Fig. 32.6). Approximately half of the axial load is trans-
mitted through the anterior channel, and half through the
two posterior channels.
Typical cervical vertebrae
The cardinal elements of a typical cervical vertebra are
its vertebral body and two articular pillars (Figs 32.1
679
and 32.2). Secondarily, transverse processes project lat-
erally from the articular pillars, and posteriorly the two
pillars are united by a pair of laminae, which support a
midline spinous process at their junction. The transverse
processes and spinous processes serve as levers upon
which act the muscles that control the position of the cer-
vical vertebrae. Along its superior, posterolateral margin
on each side, each vertebral body bears uncinate pro-
cesses. Previously enigmatic, the uncinate processes
underlie the nature of the joints between the cervical ver-
tebral bodies and how they operate.
Consecutive articular pillars are united by the zy-
gapophysial joints (Figs 32.1 and 32.2), which are synovial
joints formed by the inferior articular process of the
vertebra above and the superior articular process of
the vertebra below. Fibroadipose meniscoids intervene
between the articular cartilages of these joints (Mercer
and Bogduk, 1993). The zygapophysial joints are planar,
and at typical cervical levels are oriented at about 40° to
the coronal and transverse planes, so that they face back-
wards and upwards (Nowitzke et al., 1994). At the C2–3
level, however, the joints also face medially, such that the
pair of joints depict an ellipsoid socket into which nestles
the weight of the axis, and the load that it carries from
the head (Figs 32.2 and 32.6).
Consecutive vertebral bodies are united by interverteb-
ral discs, and by the anterior and posterior longitudinal lig-
aments (Mercer and Bogduk, 1999). The anterior ligament
connects only the typical cervical vertebrae, from C2 cau-
dally. The posterior longitudinal ligament forms a
carpet along the floor of the vertebral canal at typical cer-
vical levels, but expands into the membrana tectoria to
cover the back of the atlantoaxial region. In doing so,
the ligament separates the dural sac and spinal cord from
the mechanics of the median atlantoaxial joint.
Posterior ligaments are lacking in the cervical spine.
Interspinous ligaments are represented by only a sagittal
layer of fascia (Mercer and Bogduk, 2003). The ligamen-
tum nuchae lacks the structure of a ligament. It consists
largely of a narrow, coronal raphe, anchored to the C7
spinous process, and formed by interlacing tendons of
the splenius muscles and trapezius (Mercer and
Bogduk, 2003).
The intrinsic structure of the cervical intervertebral
discs is unlike that of lumbar discs, and differs with
age (Oda et al., 1988; Mercer and Bogduk, 1999). The
nucleus pulposus of cervical discs is gelatinous only in
children and young adults. By the age of 30 it dries
out to form a fibrocartilaginous plate (Oda et al.,
1988). Moreover, the nucleus is not surrounded by con-
centric lamellae of the anulus fibrosus (Mercer and
Bogduk, 1999). The anulus fibrosus is largely deficient
posteriorly, and consists of a thin, paramedian band of
collagen fibers that run longitudinally between the
680 N. BOGDUK

Fig. 32.7. Sketches of various views of the internal structure

of the cervical disc. In a front view, all fibers of the anterior
anulus fibrosus pass towards a point on the inferior anterior
surface of the vertebral body above. In a top view, the anulus
fibrosus (af) is crescentic in shape, thick anteriorly but tapering
at the uncinate processes (u). The nucleus pulposus (np) is a
fibrocartilaginous plate. Posteriorly the anulus is restricted
to a small bundle of paramedian, longitudinal fibers. A side
view shows the fibers of the anulus fibrosus passing upwards
and forwards. A transverse cleft runs from one uncinate pro-
cess to the other.

vertebral bodies (Fig. 32.7). Posterolaterally, the nucleus

is covered by the posterior longitudinal ligament, rather
than by anulus fibrosus. Anteriorly, the anulus fibrosus
is crescentic in shape, thin posteriorly near the uncinate
processes, but thicker anteriorly towards the midline. All
of its collagen fibers pass in a similar direction, effec-
tively aiming to a median point on the lower anterior sur-
face of the vertebral body above. This configuration
endows the anulus fibrosus with the structure of a thick
interosseous ligament that binds the anterior edges of
consecutive vertebral bodies.
The superior surface of each cervical vertebral body
presents two curvatures: a slight convex curvature along
the sagittal plane, and a deep concave curvature trans-
versely between the uncinate processes (Fig. 32.8). These
curvatures endow the vertebral body with the configura-
tion of a saddle joint (Bogduk and Mercer, 2000). Con-
sequently, the cervical interbody joints operate like a
saddle joint, with motion restricted to two planes: the
sagittal plane and an oblique coronal plane.
In the sagittal plane, the vertebral bodies can rock and
slide (rotate and translate), to provide for flexion and
extension of the neck. From above downwards, the typ-
ical cervical vertebrae exhibit progressively less transla-
tion for each degree of rotation, during flexion or
extension. This is reflected by the different locations
of their axes of movement. At higher levels the axes
Fig. 32.8. Sketches of a posterolateral view of a typical cervi-
cal vertebral body, showing the two curvatures of its superior
surface: a downward concavity along the sagittal plane, and a
second concavity, facing upwards and forwards, between the
uncinate processes (u). These two concavities endow the inter-
vertebral disc with the features of a saddle joint.
lie in the vertebral body below the moving vertebra,
but are progressively closer to their intervertebral disc
at lower levels (Amevo et al., 1991) (Fig. 32.1). These dif-
ferences correlate strongly with the height of the articu-
lar pillar at each segment (Nowitzke et al., 1994). Taller
pillars provide less space into which the vertebra can
translate once it has commenced sagittal rotation. Con-
versely, at segments with shorter pillars, sagittal rotation
lifts the inferior articular processes of the moving verte-
bra off the supporting articular pillar, and provides a
greater gap into which it can translate (Nowitzke
et al., 1994).
The second plane of movement of typical cervical
vertebrae is set at 40° forwards of the coronal plane,
and lies parallel to the plane of the zygapophysial joints
 FUNCTIONAL ANATOMY OF THE SPINE
Fig. 32.9. A sketch of a typical cervical intervertebral joint,
illustrating the mechanics of rotation of the vertebral body,
across the oblique concavity of the uncinate processes, and
around an oblique axis through the vertebral body. The motion
is like that of a cone whose apex is fixed but whose bases nev-
ertheless free to twist and spin, in the direction and plane indi-
cated by the arrow.
(Bogduk and Mercer, 2000). Across this plane, each ver-
tebra rotates like an inverted cone whose apex is fixed,
but whose base can twist (Fig. 32.9). The apex of the cone
corresponds to the anterior, median point on the verte-
bral body to which the fibers of the anulus fibrosus
are directed; and the anulus fibrosus serves to hold this
apex in place. Meanwhile the posterior, inferior edge of
the vertebral body presents a convex surface that is
cupped by the concave surface between the uncinate pro-
cesses of the vertebra below. This latter geometry is that
of an ellipsoid joint, and the posterior inferior margin of
the vertebral body is free to spin, or swing, across this
ellipsoid surface. Thus, while the anterior end of the ver-
tebral body is fixed, its posterior end and its posterior
elements are free to spin clockwise or counterclockwise
across the oblique coronal plane. During this motion, the
inferior articular processes of the zygapophysial joints
simply glide laterally across the surfaces of their sup-
porting superior articular processes.
A consequence of this mode of operation is that the
interbody joints of typical cervical vertebrae cannot tol-
erate a posterolateral anulus fibrosus, for it would
impede the spin of the posterior vertebral body across
the oblique coronal plane. Consequently, although a pos-
terior anulus is present at birth and in young children, it
gradually disappears as neck movements increase
(Tondury, 1972). By about the age of 9 years, the postero-
lateral anulus tears, and clefts appear in the region of the
uncinate processes. Progressively these clefts enlarge
centrally, until they meet in the midline, at about the
age of 30, to form a transverse cleft from one uncinate
681
Fig. 32.10. The mechanics of the early phase of whiplash
injury. As a result of a thrust from below, the cervical spine
undergoes a sigmoid deformation. Lowe segments, e.g.,
C5–6, undergo a posterior sagittal rotation around an abnor-
mally high instantaneous axis of rotation (iar), which results
in posterior elements being impacted and anterior elements
being stretched.
process to the other. This cleft is not a degenerative
change but a normal age change. The cleft effectively
forms the “joint space” across the posterior interverteb-
ral disc that allows axial rotation of the head to be accom-
modated and amplified in range by the typical cervical
vertebrae.
This structure and mechanics of the cervical spine are
of relevance to the mechanisms of injury in whiplash.
The early phase of whiplash injury involves a thrust from
below (Bogduk and Yoganandan, 2001; Bogduk, 2006).
This upward thrust deforms the cervical spine into a sig-
moid shape, within which the lower cervical vertebrae –
typically C5 and C6 – undergo an abnormal extension
(Fig. 32.10). The vertebra rocks backwards but without
translating. As a result, it rotates about an abnormally
high axis of rotation (Kaneoka et al., 1999). During this
motion, anterior elements are stretched while posterior
elements are impacted. The anterior anulus fibrosus
can be torn or avulsed, resulting in so-called rim lesions.
Impaction in the zygapophysial joints can cause impac-
tion fractures of the articular cartilages, or contusions
of the intra-articular meniscoids. During later phases,
the cervical spine rebounds into flexion, which can exces-
sively strain the capsules of the zygapophysial joints
(Curatolo et al., 2011).
Physiologic studies in laboratory animals have shown
that the capsule strains induced by whiplash injury result
in persistent nociception from the injured joint, and per-
sistent changes within the central nervous system charac-
teristic of chronic pain (Winkelstein, 2011). Clinical
studies have shown that the cervical zygapophysial joints
682 N. BOGDUK
are the single most common source of chronic neck
pain after whiplash, accounting for between 50% and
60% of cases (Bogduk, 2011). Most commonly, neck
pain – with referred pain to the shoulder girdle – stems
from the C5–6 joint, while headache stems from the
C2–3 zygapophysial joint.
Less well understood is pain from the cervical interver-
tebral discs. Conspicuously, degenerative disc disease is
not associated with neck pain. Furthermore, discogenic
pain appears to be uncommon, once zygapophysial joint
pain is taken into account (Yin and Bogduk, 2008). Per-
haps discogenic pain is caused by strains of the inteross-
eous ligament formed by the anterior anulus fibrosus, but
diagnostic techniques by which to test this proposition
have not been developed.
LUMBAR SPINE
The cardinal role of the lumbar spine is to support the
thorax and upper limbs – and any loads that they
carry – and to transmit those loads to the pelvis and lower
limbs (Bogduk, 2012a). Secondarily, the lumbar spine
accommodates a modest range of movement between
the thorax and pelvis.
In order to subserve these functions, the essential ele-
ments of the lumbar spine are the vertebral bodies of the
five lumbar vertebrae (Fig. 32.11). These are stacked into
a strong column, and are united by intervertebral discs
Fig. 32.11. Sagittal magnetic resonance images of the lumbar spine. (A) Median scan showing the vertebral bodies and spinous
processes (sp). The white dots mark the location of the axes of rotation of the vertebra above. (B) Lateral scan through the inter-
vertebral foramina and the L3–4 to L5–S1 zygapophysial joints. ped, pedicle of L3; sap, superior articular process of L4; iap,
inferior articular process of L5.
and by the anterior and posterior longitudinal ligaments
(Bogduk, 2012a). Bowing the column into a lordosis
endows the lumbar spine with the ability to absorb
dynamic axial loads (bouncing). Axial impulses deform
the lordotic curve; the energy is absorbed by the elastic
discs and longitudinal ligaments; and is returned to
restore the more upright curve, once the axial impulse
has passed (Bogduk, 2012a).
The lumbar intervertebral discs are well designed to
accommodate compression loads (Hickey and Hukins,
1980). Each consists of hydrated nucleus pulposus, sur-
rounded by an anulus fibrosus, and capped superiorly
and inferiorly by a vertebral endplate that joins the disc
to the adjacent vertebral body (Fig. 32.12). The anulus
fibrosus is formed by concentric layers of collagen
fibers, in which the fibers in any one layer run in parallel,
at about 60° to the long axis of the spine, but in succes-
sive layers that orientation alternates.
Axial compression is resisted primarily by the concen-
tric layers of the anulus fibrosus (Markolf and Morris,
1974) (Fig. 32.12). However, the tendency of the anulus
under load is to buckle, both outwards and inwards. This
buckling is resisted by the hydrostatic nucleus pulposus.
When the nucleus is compressed it exerts a radial pressure
that braces, and stiffens, the anulus, thereby preventing it
from buckling. A small range of flexion-extension is
accommodated by the discs (about 13° per segment), dur-
ing which the anulus fibrosus on the side to which
 FUNCTIONAL ANATOMY OF THE SPINE
Fig. 32.12. Close-up views of a sagittal magnetic resonance image of an L3–4 intervertebral disc. (A) The components of the disc.
np, nucleus pulposus; af, anulus fibrosus; vep, vertebral endplate. (B) The mechanics of the disc. Axial compression loads are
primarily borne by the lamellae of collagen in the anulus fibrosus. When compressed, the nucleus pulposus exerts radial pressure
to brace the anulus, and prevent it from buckling under load.
movement occurs is compressed slightly, while the anulus
on the opposite side is stretched (Bogduk, 2012a).
While strongly designed to resist compression, the
lumbar discs are poorly designed to resist axial rotation.
Because the collagen fibers of the anulus fibrosus alter-
nate in direction in successive layers, only half are avail-
able to resist axial rotation in one direction or the other.
For stability in axial rotation, the lumbar vertebral bodies
and intervertebral discs rely on the posterior elements of
the lumbar vertebrae (Bogduk, 2012a).
The posterior elements are based on an arch (Bogduk,
2012a) (Fig. 32.13). The arch is supported by stout pedi-
cles that emanate from the upper posterior surface of
each vertebral body. The pedicles serve to transmit
forces from the succeeding posterior elements to the ver-
tebral bodies, which control the position or movements
of the vertebral bodies. The arch is completed by left
and right laminae that join in the midline. From the junc-
tion of the two laminae springs a large spinous process,
and from the junction between the pedicle and lamina on
each side arises a long transverse process. These pro-
cesses serve as levers to which attach the muscles that
control the movements of the lumbar vertebrae.
At its superior and inferior lateral corners respec-
tively, each lamina bears a superior and inferior articular
process. Like large mittens, the paired superior articular
processes reach cranially to grasp the inferior articular
processes of the vertebra above, and form the zygapo-
physial joints. The plane of these joints is parallel to
the longitudinal axis of the lumbar spine. Consequently,
during flexion of the vertebral bodies, the inferior artic-
ular processes glide freely out of the sockets formed by
the superior articular processes, until movement is
arrested by tension in the joint capsules (Bogduk,
2012a). The axis of this movement typically lies in the
disc below the moving vertebra (Pearcy and Bogduk,
1988) (Fig. 32.11A), which indicates only a small amount
of translation for every degree of rotation of the moving
683
vertebra. As the inferior articular processes move, they
lift away from the superior articular process, tanta-
mount to partially subluxating the joint. Fibroadipose
meniscoids protect the exposed surfaces of the articular
cartilages during this displacement (Engel and Bogduk,
1982; Bogduk and Engel, 1984).
In axial views, the lumbar zygapophysial joints vari-
ously present flat, C-shaped, or J-shaped appearances,
which correspond to the primary functions of these
joints (Horwitz and Smith, 1940). Flat joints essentially
face medially and posteriorly. C-shaped joints have an
anterior end that faces posteriorly, and a posterior end
that faces medially. J-shaped joints have a small anterior
lip facing posteriorly, and a larger surface facing medi-
ally. The medially facing surfaces serve to resist axial
rotation of the vertebrae. Attempted axial rotation
swings the inferior articular process laterally, but this
movement is arrested by the opposing superior articular
process. The range of motion is limited to about 2° or less
per segment (Pearcy and Tibrewal, 1984), and is accom-
modated only by compression of the articular cartilage.
The surfaces that face posteriorly serve to resist forward
displacement (listhesis) of the vertebra.
Impaction of an inferior articular process against its
superior articular process tends to force the inferior pro-
cess backwards, and lift the lamina from which it arises
(like opening a hatchback). In turn this tendency stresses
the junction between the lamina and its pedicle. Repeated
impactions – particularly during repeated axial
rotation – can cause stress fractures at this point, result-
ing in pars interarticularis defects.
The lumbar zygapophysial joints can be a source of
low-back pain, but its prevalence is uncertain. It appears
to be uncommon or rare in injured workers, but is com-
mon in elderly patients (Bogduk, 2008, 2012b).
The most common cause of chronic low-back pain is
internal disc disruption (Bogduk et al., 2013). This condi-
tion is characterized by degradation of the nucleus
684 N. BOGDUK

Fig. 32.13. Sketches of the posterior elements of a lumbar vertebra. (A) Posterior view. The two laminae (la) form a quadrangular
plate, from whose corners project the superior (sap) and inferior (iap) processes. From the junction of the two laminae projects the
spinous process (sp). On each side, the inferior and superior articular processes of consecutive vertebrae form the zygapophysial
joint (zj) (B) Axial (top) view. The posterior elements are connected to the vertebral body (vb) by the pedicles (p). The transverse
process (tp) projects from the junction of the pedicle and lamina, on each side.

pulposus of the affected disc and the development of
radial fissures into the posterior or posterolateral anulus.
The condition has been produced in laboratory animals,
and pursued in numerous clinical studies. Its cause is
compression injuries that produce small fractures of
the vertebral endplate. These result in degradation of
the matrix of the nucleus pulposus. As the nucleus
becomes less able to retain water, it is no longer able
to pressurize and brace the anulus. Pressures in the
nucleus drop, but rise in the posterior anulus. The
unbraced anulus progressively delaminates, particularly
in regions of high stress where the laminae are curved: at
the posterolateral corners or the posterior paramedian
sector. Pain arises as a result of chemical irritation of
nociceptors in the anulus by degradation products from
the nucleus, and as a result of the increased mechanical
stresses on the surviving, intact laminae of anulus
(Bogduk et al., 2013). To various degrees of certainty
the condition can be diagnosed by characteristic features
on magnetic resonance imaging, such as Modic lesions in
the vertebral body or high-intensity zones in the anulus
fibrosus, and by provocation discography (Bogduk
et al., 2013). No treatment has been vindicated, but sev-
eral minimally invasive interventions are being pursued,
which encompass ablating nociceptors in the disc, inject-
ing restorative agents such as stem cells, or injecting
antagonists of inflammation.
MUSCLES
The anatomy of muscles of the cervical and lumbar spine
is made complex by the diversity of their numerous
attachments. If those specifics are ignored, the anatomy
becomes simpler.
Small muscles connect consecutive spinous processes
and transverse processes. Too small to move their verte-
brae effectively, these muscles serve as proprioceptors
for the spine (Bastide et al., 1989).
Prevertebral muscles are represented only in the cer-
vical spine (Standring, 2008). The longus cervicis con-
nects the vertebral bodies and transverse processes of
the cervical vertebrae. It is covered by the longus capitis
which anchors the skull to the cervical vertebrae. These
muscles are weak flexors of the head and neck.
Various suboccipital muscles control movements
of the head in relation to the atlas and the axis. They
are the rectus anterior and rectus lateralis anteriorly,
and the rectus capitis posterior major and minor acc-
ompanied by obliquus inferior and obliquus superior,
posteriorly (Standring, 2008). Collectively these muscles
control the orientation of the head on the atlas and axis.
The postvertebral muscles are aligned systematically,
side by side and by layers (Standring, 2008). Multifidus
is the deepest and most medial muscle. Its fascicles arise
from a spinous process and descend to various insertions
on articular processes and transverses processes one to
several segments caudally. It is flanked by the longissi-
mus system of muscles, which attach to transverse pro-
cesses near their bases, and whose components are large
at lumbar levels, but virtually miniscule at cervical levels.
Further laterally runs the iliocostalis system, which
attaches to transverse processes near their tips, and
whose components are, likewise, large at lumbar levels
but miniscule at cervical levels. A semispinalis system
is vestigial at lumbar levels but well developed at cervical
levels. Semispinalis cervicis arises from the cervical spi-
nous processes, and covers the multifidus with fascicles
longer than those of the latter muscle. Semispinalis capi-
tis arises from the occiput, and is anchored to the cervical
 FUNCTIONAL ANATOMY OF THE SPINE 685
transverse processes. It is the largest of the posterior lateral atlantoaxial joint before joining the cervical
neck muscles. It is covered by the splenius muscle, which plexus (Lazorthes and Gaubert, 1956).
passes cranially and laterally from the raphe of the liga- The C3–7 cervical spinal nerves lie above their like-
mentum nuchae to wrap around all the other posterior numbered vertebrae, enclosed in their respective inter-
muscles of the neck. Splenius cervicis reaches the upper vertebral foramina. They are joined by the C8 spinal
cervical transverse processes, while splenius capitis nerve, which lies in the C7–T1 intervertebral foramen.
reaches the superior nuchal line. Variously and collec- The ventral rami of C3 and C4 join the cervical plexus,
tively, the cervical postvertebral muscles act to extend and the lower cervical ventral rami join the brachial
the head and the cervical spine. plexus. The dorsal rami of the typical cervical spinal
Other muscles use the vertebral column adventi- nerves form lateral branches that supply the splenius,
tiously, as a base from which to act on nonspinal struc- longissimus, and iliocostalis; and medial branches that
tures. In the neck, these include the scalene muscles, supply the deeper and medial posterior neck muscles,
which act on the ribs; and levator scapulae and trapezius, and the cervical zygapophysial joints (Bogduk, 1982).
which act on the shoulder girdle. Sternocleidomastoid is The cervical medial branches have constant locations
the principal flexor and rotator of the head and neck, but on the cervical articular pillars, which allow them to be
passes directly from the manubrium and clavicle to the targeted for fluoroscopy-guided diagnostic blocks, by
head, with no connection to the cervical spine. Being which pain from the zygapophysial joints can be diag-
locked to the skull through the atlanto-occipital joints, nosed (Bogduk, 1982, 2011).
the atlas is rotated when the sternocleidomastoid rotates Gray rami communicantes, from the stellate ganglion
the head. and from the cervical ventral rami, form a plexus –
In the lumbar spine, psoas major arises from the ver- called the vertebral nerve – that accompanies the verte-
tebral bodies, discs, and transverse processes to act on bral artery through the foramina transversaria of
the femur, but does not move the lumbar spine the neck, and into the posterior cranial fossa (Bogduk
(Bogduk et al., 1992). Quadratus lumborum attaches et al., 1981a). Although migraine cervicale, or the
to the lumbar transverse processes but acts principally Barré–Lieou syndrome, has been attributed to irritation
on the 12th rib; its actions on the lumbar vertebrae are of these nerves, and spasm of the vertebral artery, labo-
effectively trivial (Phillips et al., 2008). Transversus ratory studies have shown the vertebrobasilar system to
abdominis stems from the lumbar transverse processes, be remarkable unresponsive to stimulation of the verte-
and has virtually no effect on the lumbar vertebrae bral nerve (Bogduk et al., 1981a; Lambert et al., 1984).
(Macintosh et al., 1987). Likewise, latissimus dorsi gains The cervical sinuvertebral nerves are formed by
some anchorage to the lumbar spinous processes but has somatic roots from the ventral rami and autonomic roots
a negligible action on the lumbar spine (Bogduk from the rami communicantes in the vertebral nerve. As
et al., 1998). recurrent meningeal branches they innervate the cervical
dural sac, but also innervate the cervical discs and the
posterior longitudinal ligament (Bogduk et al., 1988).
The C1–3 sinuvertebral nerves innervate the ligaments
INNERVATION
of the median atlantoaxial joint before passing through
The C1 spinal nerve is unlike other spinal nerves, which foramen magnum to supply the dura mater over the cli-
reinforces the atlas being suboccipital rather than cervi- vus (Kimmel, 1960).
cal in nature. This nerve lacks a typical dorsal root gan- The lumbar spinal nerves lie obliquely in their inter-
glion, but ganglion cells can be found amongst the vertebral foramina, each below the like-numbered verte-
rootlets of the spinal accessory nerve. The C1 dorsal bra. Their ventral rami enter the lumbar or lumbosacral
ramus appears amongst the posterior suboccipital mus- plexus. Their dorsal rami form lateral and intermediate
cles (Lazorthes and Gaubert, 1956). Sometimes it can branches that innervate the iliocostalis and longissimus
have a cutaneous branch. The C1 ventral ramus crosses muscles respectively (Bogduk et al., 1982; Bogduk,
the posterior arch of the atlas, behind the superior artic- 1983). Medial branches innervate the lumbar zygapophy-
ular process. It innervates the atlanto-occipital joint sial joints and the multifidus (Bogduk et al., 1982;
before entering the cervical plexus (Lazorthes and Bogduk, 1983). Where the medial branches cross the root
Gaubert, 1956). of the superior articular process they can be targeted for
The C2 spinal nerve lies behind the lateral atlantoaxial fluoroscopy-guided diagnostic blocks, by which pain
joint, and forms a large dorsal ramus that supplies the from the lumbar zygapophysial joints can be diagnosed
more superficial posterior neck muscles, and becomes (Bogduk, 1983, 2008, 2012b).
cutaneous as the greater occipital nerve, over the occiput At each segmental level, the lumbar sinuvertebral
(Bogduk, 1982). The C2 ventral ramus supplies the nerves arise from the ventral ramus and gray ramus
686 N. BOGDUK
Fig. 32.14. Sagittal magnetic resonance images of the thoracic
spine. (A) Median section, through the vertebral bodies, spinal
cord (sc), and spinous processes (sp) cord. (B) Paramedian sec-
tion through the zygapophysial joints (zj). Intervertebral discs
(ivd) are evidence in both sections. (Courtesy of Dr. Tim Maus,
Mayo Clinic, Rochester MN.)
communicans. Each passes back into the intervertebral
foramen to supply the dural sac, the posterior longitudi-
nal ligament, and the posterior anulus fibrosus (Bogduk
et al., 1981b; Bogduk, 1983). These nerves provide the
sensory pathway for lumbar discogenic pain.
THORACIC SPINE
There have been no substantial advances in the descrip-
tion of the anatomy of the thoracic spine since editions
of anatomy textbooks of the 19th and 18th century. In
parallel, there has been little advance in the understand-
ing of thoracic spinal pain and its sources, let alone
causes. No diagnostic or treatment procedures have been
validated. Thoracic spinal pain essentially remains a
mystery.
Like cervical and lumbar vertebrae, the thoracic ver-
tebrae have vertebral bodies that are connected by inter-
vertebral discs and longitudinal ligaments, and posterior
elements that are connected by zygapophysial joints
(Fig. 32.14). The distinction of the thoracic spine is that
it suspends the ribs. At typical thoracic levels, the head
of the rib articulates with the intervertebral disc and
demifacets on the edges of the vertebrae that bind that
disc, and the articular tubercle of the rib articulates with
the transverse process of the upper of the two vertebrae
Fig. 32.15. Axial magnetic resonance image of a typical tho-
racic spinal segment. vb, vertebral body; zj, zygapophysial
joint; sp, spinous process; cvj, costovertebral joint; ctj,
costotransverse joint. (Courtesy of Dr. Tim Maus, Mayo
Clinic, Rochester, MN.)
(Fig. 32.15). Exceptions to this arrangement occur at T1
and at T11 and T12, where the head of the rib fully artic-
ulates with the like-numbered vertebrae.
Few studies have explored the innervation of the tho-
racic spine (Bogduk, 2002). The thoracic sinuvertebral
nerves are assumed to be homologous to those at cervical
or lumbar levels. The courses of the thoracic dorsal rami
appear to differ from those at cervical and lumbar levels,
but are nevertheless homologous (Chua and Bogduk,
1995). Whereas the medial branches at cervical and lum-
bar levels wind around the base of the superior articular
process at each segmental level, at thoracic levels the
dorsal ramus stretches to the tip of the transverse pro-
cess before dividing into medial and lateral branches.
This difference is reconciled once it is realized that what
are called the transverse processes at cervical and lumbar
levels are embryologically costal elements (rudimentary
ribs), whereas the embryologic transverse elements (or
true transverse processes) are absorbed into the base
of the superior articular process. Consequently, at cervi-
cal and lumbar levels, the medial branches cross the
superior articular process because the true transverse
processes also lie there. This distinction becomes perti-
nent for minimally invasive, diagnostic, and treatment
procedures that target thoracic medial branches. The tar-
get lies on the transverse process, not on the superior
articular process (Chua and Bogduk, 1995).
A persisting curiosity pertains to the structure of tho-
racic intervertebral discs. Cervical discs differ greatly
from lumbar discs, but undiscovered is the transition
 FUNCTIONAL ANATOMY OF THE SPINE
zone. Are thoracic discs like cervical discs, or do they have
the structure of lumbar discs? Given that cervical uncinate
processes are homologous to the heads of the ribs, unpub-
lished observations suggest that discs change their struc-
ture where uncinate processes or their rib equivalent
cease. Thoracic discs become lumbar in nature at T11,
where the rib no longer articulates with the disc.
REFERENCES
Amevo B, Worth D, Bogduk N (1991). Instantaneous axes of
rotation of the typical cervical motion segments: a study in
normal volunteers. Clin Biomech 6: 111–117.
Bastide G, Zadeh J, Lefebvre D (1989). Are the ‘little muscles’
what we think they are? Surg Radiol Anat 11: 255–256.
Bogduk N (1982). The clinical anatomy of the cervical dorsal
rami. Spine 7: 319–330.
Bogduk N (1983). The innervation of the lumbar spine. Spine
8: 286–293.
Bogduk N (2002). Innervation and pain patterns of the thoracic
spine. In: R Grant (Ed.), Physical therapy of the Cervical
and Thoracic Spine, 3rd edn. Churchill Livingstone, New
York, pp. 73–81.
Bogduk N (2006). Whiplash injury. In: F Cervero, TS Jensen
(Eds.), Handbook of Clinical Neurology Vol. 81: Pain,
Elsevier, Amsterdam, pp. 791–801.
Bogduk N (2008). Evidence-informed management of chronic
back pain with facet injections and radiofrequency neurot-
omy. Spine J 8: 56–64.
Bogduk N (2011). On cervical zygapophysial joint pain after
whiplash. Spine 36: S194–S199.
Bogduk N (2012a). Clinical Anatomy of the Lumbar Spine and
Sacrum, 5th edn. Elsevier, Edinburgh.
Bogduk N (2012b). Lumbar medial branch neurotomy. In:
S Dagenais, S Haldeman (Eds.), Evidence-Based
Management of Low Back Pain, Elsevier, St Louis,
pp. 351–363.
Bogduk N (2014). The neck and headaches. Neurol Clin 32:
471–487.
Bogduk N, Bartsch T (2008). Cervicogenic headache. In:
SD Silberstein, RB Lipton, DW Dodick (Eds.), Wolff’s
Headache, 8th edn. Oxford University Press, New York,
pp. 551–570.
Bogduk N, Engel R (1984). The menisci of the lumbar zyga-
pophyseal joints. A review of their anatomy and clinical
significance. Spine 9: 454–460.
Bogduk N, Govind J (2009). Cervicogenic headache: an
assessment of the evidence on clinical diagnosis, invasive
tests, and treatment. Lancet Neurol 8: 959–968.
Bogduk N, Mercer SR (2000). Biomechanics of the cer-
vical spine. I: Normal Kinematics. Clin Biomech 15:
633–648.
Bogduk N, Yoganandan N (2001). Biomechanics of the cervi-
cal spine Part 3: minor injuries. Clin Biomech 16: 267–275.
Bogduk N, Lambert G, Duckworth JW (1981a). The anatomy
and physiology of the vertebral nerve in relation to cervical
migraine. Cephalalgia 1: 11–24.
Bogduk N, Tynan W, Wilson AS (1981b). The nerve supply to
the human lumbar intervertebral discs. J Anat 132: 39–56.
687
Bogduk N, Wilson AS, Tynan W (1982). The human lumbar
dorsal rami. J Anat 134: 383–397.
Bogduk N, Windsor M, Inglis A (1988). The innervation of the
cervical intervertebral discs. Spine 13: 2–8.
Bogduk N, Pearcy M, Hadfield G (1992). Anatomy and biome-
chanics of psoas major. Clin Biomech 7: 109–119.
Bogduk N, Johnson G, Spalding D (1998). The morphology
and biomechanics of latissimus dorsi. Clin Biomech 13:
377–385.
Bogduk N, Aprill C, Derby R (2013). Lumbar discogenic pain:
state-of-the-art review. Pain Med 14: 813–836.
Chua WH, Bogduk N (1995). The surgical anatomy of thoracic
facet denervation. Acta Neurochir 136: 140–144.
Curatolo M, Bogduk N, Ivancic PC et al. (2011). The role of
tissue damage in whiplash-associated disorders. Spine 36:
S309–S315.
Dvorak J, Hayek J, Zehnder R (1987). CT-functional diagnos-
tics of the rotatory instability of the upper cervical spine
part 2. An evaluation on healthy adults and patients with
suspected instability. Spine 12: 726–731.
Engel R, Bogduk N (1982). The menisci of the lumbar zyga-
pophysial joints. J Anat 135: 795–809.
Fielding JW, Cochran G van B, Lawsing JF et al. (1974). Tears
of the transverse ligament of the atlas. J Bone Joint Surg
56A: 1683–1691.
Hickey DS, Hukins DWL (1980). Relation between the struc-
ture of the anulus fibrosus and the function and failure of
the intervertebral disc. Spine 5: 100–116.
Horwitz T, Smith RM (1940). An anatomical, pathological
and roentgenological study of the intervertebral joints
of the lumbar spine and of the sacroiliac joints. Am
J Roentgenol 43: 173–186.
Kaneoka K, Ono K, Inami S et al. (1999). Motion analysis of
cervical vertebrae during whiplash loading. Spine 24:
763–770.
Kimmel DL (1960). Innervation of the spinal dura mater and
dura mater of the posterior cranial fossa. Neurology 10:
800–809.
Koebke J, Brade H (1982). Morphological and functional stud-
ies on the lateral joints of the first and second cervical ver-
tebrae in man. Anat Embryol 164: 265–275.
Lambert GA, Duckworth JW, Bogduk N et al. (1984). Low
pharmacological responsiveness of the vertebro-basilar cir-
culation in Macaca nemestrina monkeys. Eur J Pharmacol
102: 451–458.
Lazorthes G, Gaubert J (1956). L’innervation des articulations
interapophysaire vertebrales. Comptes Rendues de
l’Association des Anatomistes 43: 488–494.
Macintosh JE, Bogduk N, Gracovetsky S (1987). The biome-
chanics of the thoracolumbar fascia. Clin Biomech 2:
78–83.
Markolf KL, Morris JM (1974). The structural components
of the intervertebral disc. J Bone Joint Surg 56A:
675–687.
Mercer S, Bogduk N (1993). Intra-articular inclusions of the
cervical synovial joints. Br J Rheumatol 32: 705–710.
Mercer S, Bogduk N (1999). The ligaments and anulus fibro-
sus of human adult cervical intervertebral discs. Spine 24:
619–626.
688 N. BOGDUK
Mercer SR, Bogduk N (2003). Clinical anatomy of ligamen-
tum nuchae. Clin Anat 16: 484–493.
Nowitzke A, Westaway M, Bogduk N (1994). Cervical zyga-
pophyseal joints: geometrical parameters and relationship
to cervical kinematics. Clin Biomech 9: 342–348.
Oda J, Tanaka H, Tsuzuki N (1988). Intervertebral disc
changes with aging of human cervical vertebra from the
neonate to the eighties. Spine 13: 1205–1211.
Pearcy MJ, Bogduk N (1988). Instantaneous axes of rotation of
the lumbar intervertebral joints. Spine 13: 1033–1041.
Pearcy MJ, Tibrewal SB (1984). Axial rotation and lateral
bending in the normal lumbar spine measured by three-
dimensional radiography. Spine 9: 582–587.
Phillips S, Mercer S, Bogduk N (2008). Anatomy and
biomechanics of quadratus lumborum. J Eng Med 222:
151–159.
Standring S (Ed.), (2008). Gray’s Anatomy, 40th edn.
Churchill Livingstone, Edinburgh, pp. 736–743.
Tondury G (1972). The behaviour of the cervical discs during
life. In: C Hirsch, Y Zotterman (Eds.), Cervical pain,
Pergamon Press, Oxford, pp. 59–66.
Wasserman BR, Moskovitch R, Razi AE (2011). Rheumatoid
arthritis of the cervical spine. Clinical considerations. Bull
Hosp Joint Dis 68: 136–148.
Winkelstein BA (2011). How can animal models inform on the
transition to chronic symptoms in whiplash? Spine 36:
S218–S225.
Wortzman G, Dewar FP (1968). Rotary fixation of the atlan-
toaxial joint: rotational atlantoaxial subluxation.
Radiology 90: 479–487.
Yin W, Bogduk N (2008). The nature of neck pain in a private
pain clinic in the United States. Pain Med 9: 196–203.
Handbook of Clinical Neurology, Vol. 136 (3rd series)
Neuroimaging, Part II
J.C. Masdeu and R.G. González, Editors
© 2016 Elsevier B.V. All rights reserved

Chapter 33

Neuroimaging of spine tumors

NANDOR K. PINTER, THOMAS J. PFIFFNER, AND LASZLO L. MECHTLER*
Dent Neurologic Institute, Amherst, NY, USA

Abstract
Intramedullary, intradural/extramedullary, and extradural spine tumors comprise a wide range of neo-
plasms with an even wider range of clinical symptoms and prognostic features. Magnetic resonance imag-
ing (MRI), commonly used to evaluate the spine in patients presenting with pain, can further characterize
lesions that may be encountered on other imaging studies, such as bone scintigraphy or computed tomog-
raphy (CT). The advantage of the MRI is its multiplane capabilities, superior contrast agent resolution,
and flexible protocols that play an important role in assessing tumor location, extent in directing biopsy, in
planning proper therapy, and in evaluating therapeutic results. A multimodality approach can be used to
fully characterize the lesion and the combination of information obtained from the different modalities
usually narrows the diagnostic possibilities significantly. The diagnosis of spinal tumors is based on
patient age, topographic features of the tumor, and lesion pattern, as seen at CT and MRI. The shift
to high-end imaging incorporating diffusion-weighted imaging, diffusion tensor imaging, magnetic res-
onance spectroscopy, whole-body short tau inversion recovery, positron emission tomography, intrao-
perative and high-field MRI as part of the mainstream clinical imaging protocol has provided
neurologists, neuro-oncologists, and neurosurgeons a window of opportunity to assess the biologic behav-
ior of spine neoplasms. This chapter reviews neuroimaging of spine tumors, primary and secondary,
discussing routine and newer modalities that can reduce the significant morbidity associated with these
neoplasms.

INTRODUCTION advances in the care of patients with spine tumors.

The characterization of spine tumors by MRI involves
The historic classification of spine tumors is based on
determining, in the context of patient’s age and sex,
the use of myelography with three main groups, as sche-
the location of the lesion and whether or not it enhances
matically depicted in Figure 33.1: (1) extradural extrame-
after gadolinium injection. Computed tomography (CT)
dullary; (2) intradural extramedullary; and (3) intradural
best delineates osseous integrity while MRI is better
intramedullary. The incidence of metastatic disease
at assessing soft-tissue involvement. The purpose of
involving the vertebrae, epidural space, and leptome- this chapter is to describe the neuroimaging findings
ninges accounts for 97% of tumors involving the
of spine tumors based on the location of the tumor in
spine. Primary tumors of the spine, spinal cord, spinal
its relationship to the dura and spinal cord (Mechtler
meninges, and cauda equina are relatively rare (Duong
and Nandigam, 2013).
et al., 2012). Data from national registries and improved
imaging capabilities have allowed spine tumor specialists
METASTATIC TUMORS OF THE
the opportunity to study and treat these unusual and rare
VERTEBRAL COLUMN
tumors with more confidence and better results. The
introduction of magnetic resonance imaging (MRI) to The spine is the third most common site for metastatic
clinical practice has been one of the most important disease and the most common site for bone metastasis

*Correspondence to: Laszlo L. Mechtler, MD, Professor of Neurology and Oncology, Dent Neurologic Institute, 3980 Sheridan
Drive, Amherst NY 14226, USA. Tel: +1-716-250-2000, Fax: +1-716-250-2045, E-mail: lmechtler@dentinstitute.com
690 N.K. PINTER ET AL.
Fig. 33.1. Historic classification of spine tumors based on computed tomography myelography. (A) Normal, (B) extradural extra-
medullary, (C) intradural extramedullary, and (D) intradural intramedullary. From Mechtler and Nandigam (2013).
(Shah and Salzman, 2011). Metastatic disease of the ver-
tebral column is more frequent than primary neoplastic
diseases. Approximately two-thirds of cancer patients
will develop bone metastasis and symptomatic spinal
metastasis will occur in almost 10% of cancer patients.
The most common primary sites are the prostate, breast,
kidney, lung, and thyroid. The incidence of skeletal
metastases according to the primary tumor is as follows;
breast 73% (47–85%), prostate 68% (33–85%), thyroid
42% (28–60%), lung 36% (30–55%), kidney 35%
(33–40%), esophageal 6% (5–7%), and gastrointestinal
5% (3–11%) (Maccauro et al., 2011) The most common
cause of metastatic spine disease is breast cancer in
women; however, in men, prostate cancer is most com-
mon. The thoracic spine is the most commonly involved.
The majority of the lesions are extradural in location,
consisting of lesions which are localized to the epidural
space and those which are nested in the vertebral body.
Prostate, breast, and lung cancer are again the leading
cause of spinal cord compression, each accounting for
about 15–20% of the cases. The remaining cancers stem
from renal cell, non-Hodgkin’s lymphoma, multiple
myeloma, colorectal cancers, sarcomas, and unknown
tumors. Pain, the most common initial feature, occurs
in 95% of adults and 80% of children. Pain is usually
localized to the site of metastasis and is caused by
stretching the pain-sensitive bony periosteum. Radicular
pain is less frequent but is also localizing. Nocturnal pain
upon lying down is typical.
Three types of bone metastasis are distinguished:
osteolytic, osteoblastic, and mixed; 71% are osteolytic,
8% are osteoblastic, and 21% are mixed.
Osteolytic metastases typically develop in cancers
of the breast, lung, kidney, thyroid, oropharyngeal can-
cers (Shah and Salzman, 2011) and in melanoma (Sun
et al., 2013). This is a result of osteoclast activation, rather
than a direct invasion of bone tissue by tumor cells.
In osteoblastic metastases the balance of bone meta-
bolism is shifted to the benefit of bone production as
a result of pathologic activation of osteoblasts. Osteo-
blastic lesions usually occur in prostate, bladder and
nasopharyngeal cancer, medulloblastoma, neuroblasto-
mas, and bronchial carcinoid (Long et al., 2010).
Imaging of vertebral metastases
In today’s clinical practice MRI is the most important
modality in imaging of metastatic spine disease. Plain
film is no longer the routine diagnostic toolbar due to
its low sensitivity and specificity (Salvo et al., 2009;
Shah and Salzman, 2011). Nuclear medicine studies have
a well-defined role in metastasis imaging. Bone scans
have been used for screening, since the tracer accumu-
lates in metastatic sites with high sensitivity, thus reflect-
ing the increased bone turnover. The sensitivity and
specificity of bone scans were improved with single-
photon emission computed tomography (SPECT) scans
(Ryan and Fogelman, 1995). Flurodeoxyglucose (F18-
FDG) positron emission tomography (PET) alone and
PET CT can discover spinal metastases with a sensitivity
of 74% and 98%, respectively (Metser et al., 2004). F18-
FDG PET has been reported to be more sensitive in
detecting osteolytic metastases (Cook and Fogelman,
2000). CT has a lower sensitivity in detecting osseous
metastases and an inferior diagnostic accuracy com-
pared to MRI (Buhmann Kirchhoff et al., 2009). In fact,
CT is less accurate in detecting paraspinal soft tissue,
bone edema, and bone metastases that may be missed
if destruction is not present (Shah and Salzman, 2011).
Therefore, CT has a rather complementary role in
first-line imaging of spinal metastases and owns priority
only in those cases when the integrity and fine structure
of the trabecular and cortical bone are a question, preop-
erative planning is required, or when MRI is contraindi-
cated. MRI is superior to CT in all other cases.
Metastatic lesions are most commonly focal or multi-
focal and the diffuse involvement of the vertebral bodies
is less common. Focal abnormalities are hypointense
on T1 and hyperintense on T2 and short tau inversion
recovery (STIR) sequences. In general, metastases will
enhance with contrast, although it is important to always
acquire a noncontrast study for comparison. The diffuse
 NEUROIMAGING OF SPINE TUMORS 691

Fig. 33.2. Diffuse metastatic involvement of the cervical and upper thoracic spine. The T1-weighted image (A) shows heteroge-
neous signal intensity in all vertebral bodies, with mostly isointense to hypointense signal accompanied by focal hyperintense areas
(open arrow) compared to intervertebral discs. On T2-weighted image (B), similar heterogeneity can be seen with hyperintense and
isointense areas. On short tau inversion recovery (STIR) sequence (C) diffuse hyperintense signal correlates with the T1 hypoin-
tensity, typical of metastatic disease of the bone marrow. The involvement of spinous processes is highly suspicious on T2 and
becomes unquestionable on STIR (arrows). The focal hyperintensities on T1 are associated with fatty marrow and represented as
hypointense signal on STIR.

marrow involvement can be difficult to assess, because a
generally low signal intensity appearance can be mislead-
ing, giving the false impression of normal marrow. It is
helpful to compare the marrow’s signal intensity to that
of the discs and muscles. In adults it can be regarded as
abnormal if the marrow has lower signal intensity than
discs or muscles (Fig. 33.2) (Long et al., 2010). Metasta-
ses tend to occur in the posterior part of the vertebral
body, involving the pedicles. Most often metastases
are destructive and can be expansive.
In osteolytic metastases, the cancellous bone is
replaced by tumor tissue and acceleration of bone
resorption occurs due to an imbalance in metabolism.
The focal disequilibrium in calcified elements will create
the characteristic image of circumscribed translucency
on radiographs and a generally low-density area framed
by intact, partly intact, or sclerotic osseous components
on CT. On MRI hypointense T1, hyperintense T2, and
STIR signals will represent the structural and biochem-
ical (e.g., edema) changes in the bone marrow (Long
et al., 2010). Posterior cortical and pedicle destruction
is not infrequent and best visualized by CT. During
the course of antitumor therapy progressive sclerosis
may be visible on follow-up imaging, indicating positive
response to therapy. Fluid levels may occur in lytic
metastases (Jarraya et al., 2013) and might even mimic
primary tumors (Colangeli et al., 2010). The term osteo-
blastic refers to the biologic behavior of the lesions, but
from the imaging point of view, sclerotic or osteosclero-
tic may sound more practical, as they reflect the imaging
findings. Osteoblastic metastases are usually focal or
mottled and in most cases are multifocal. Diffuse
involvement of the vertebral body occurs occasionally.
On MRI the lesions show T1 and T2 hypointensity. On
T2 and STIR images a bright rim surrounding the scle-
rotic lesions may be depicted, reflecting bone marrow
edema. This is known as the “halo” sign. Edematous sig-
nal can often be observed on STIR, expanding from the
body to the pedicles, without a well-defined lesion on T1
or T2 sequences. The enhancement pattern in MRI may
vary depending on the degree of sclerosis. On CT and
plain film these lesions show high density resulting from
excessive calcification. The epidural expansion of the
tumor may create the “draped curtain sign,” which can
be depicted on axial MRI slices (Fig. 33.3).
Vertebral compression deformity
Vertebral compression deformity is frequent in elderly
patients. The cause of this abnormality can be benign
or malignant. Distinguishing between benign and patho-
logic vertebral body compression fractures is usually
possible on MRI. Chronic benign fractures typically
have marrow signal intensity that is isointense with nor-
mal vertebrae on all sequences. In addition, there is
no involvement of the posterior elements, absence of
paravertebral or epidural mass, and preservation of pos-
terior cortex. Pathologic fractures show comparatively
low signal intensity on T1- and high signal intensity
on T2-weighted sequences (Fig. 33.4) (Griffith and
Guglielmi, 2010). Pathologic compression fractures
typically enhance with contrast; however, conventional
MR techniques cannot always be used to differentiate
benign from malignant lesions due to their similar
appearances. For example, osteoporotic compression
fracture can be confused with metastatic compression
692 N.K. PINTER ET AL.

Fig. 33.3. A 48-year-old man with multiple myeloma. Sagittal (A) and axial (B) contrast-enhanced T1-weighted scans with fat-
suppression shows a large enhancing mass occupying the T12 vertebral body, extending into the spinal canal and the left pedicle
(arrows). The yellow arrows on the axial image mark as the tumor invades the anterior epidural space bilaterally and compression
on the thecal sac, creating the “draped curtain sign.” Reformatted computed tomography in the coronal and sagittal planes
(C and D) shows the osteolytic nature of this tumor. This was thought to be a solitary plasmacytoma. Further imaging revealed
“punched-out” osteolytic lesions (black arrows) characteristic of multiple myeloma.

Fig. 33.4. Compression fracture resulting from metastatic disease. T1 hypointense (A), T2 intermediate-hyperintense (B) and
short tau inversion recovery STIR hyperintense signal (C) is demonstrated in the compressed T9 vertebral body. The body shows
a vertebral plana deformity. Posteriorly focal sclerotic changes can be observed on the sagittal reconstruction of CT scan (D).

in the acute phase. Edema in an acute benign compres-
sion fracture replaces the normal marrow, resulting in
hypointensity on T1-weighted images and hyperintensity
on T2-weighted images. The vertebral body with benign
fracture may enhance with contrast. These MR signal
intensity characteristics are similar to those of metastases
and cause ambiguity, especially when only a single lesion
is present. Pathologic fractures are often multiple; other
key features of pathologic compression fractures include
loss of the posterior body height, pedicle involvement,
epidural and paraspinal mass.
Diffusion-weighted imaging (DWI) is emerging as a
powerful clinical tool for directing the care of patients
with cancer (Padhani et al., 2011). The basic biologic pre-
mise for the use of DWI is the characteristic increased
cellular content of malignant tissue, and increased water
content. These features result in higher signal intensity
of malignant disease on high b-value images with corre-
sponding low apparent diffusion coefficient (ADC)
values (Khoo et al., 2001). This is especially helpful in
patients with neuroblastomas, leukemia, lymphoma,
rhabdomyosarcoma, Ewing sarcoma, and metastatic
primitive neuroectodernal tumors. DWI can also be a
useful tool for distinguishing acute benign osteoporotic
from malignant vertebral compression fractures, result-
ing in low or isointense signal on DWI and high ADC
values on ADC maps (Duvauferrier et al., 2013;
Rumpel et al., 2013).
Whole-body STIR has also been used to monitor dis-
ease progression in patients with lymphoma, multiple
myeloma, metastatic primitive neuroectodernal tumors,
and neurofibromatosis (Fayad et al., 2013).
PRIMARY TUMORS OF THE SPINE
Benign primary spine tumors
Hemangioma is the most common primary tumor of the
spine; the incidence from autopsy studies is estimated to
 NEUROIMAGING OF SPINE TUMORS
be 10% (Ropper et al., 2011). Histologically they consist
of multiple small, thin-walled vessels interspersed with
trabeculae and infiltrating the medullary cavity. Heman-
giomas have a characteristic, polka-dot appearance on
the axial CT images, which represent the cross-section
of the prominent vertical trabeculae. On MRI lesions
show high signal intensity on T1 and T2 images, often
with a heterogeneous structure, which may also contain
flow voids as signs of vascular elements. Osteoid
osteomas occur in young adults and their most common
symptom is a painful scoliosis that worsens at night
(Theodorou et al., 2008). They usually arise from the
posterior elements. Lumbar spine involvement is typical.
On radiographs osteoid osteomas have the appearance
of a lytic lesion surrounded by sclerotic rim. The same
finding is present on CT scans. The sclerotic changes
may affect the central area or nidus. On T1 images the
nidus displays intermediate signal intensity and can have
signal voids as a result of calcification. On T2 images the
nidus appears as a low-intensity area surrounded by
high-intensity rim representing edema. Osteoblastomas
have a nonspecific MRI appearance, with T1 and T2 pro-
longation depending on the degree of matrix mineraliza-
tion. Osteoblastomas have a reactive rim similar to
osteoid osteomas, which can lead to overestimation of
lesion (Chai and Cho, 2013).
Aneurysmal bone cysts are expansile lesions contain-
ing blood-filled cysts separated by walls of bone (Long
et al., 2010). The spine is affected in 12–30% of all aneu-
rysmal bone cyst cases. On CT scans ballooning, lobu-
lated cystic lesions can be found occasionally with
693
fluid–fluid levels (Ropper et al., 2011). The latter is indic-
ative of hemorrhage and may be more sensitively visual-
ized by MRI (Fig. 33.5). Signal changes, representing
blood degradation products of different age, can be
appreciated (Vidal and Murphey, 2007). Giant cell tumor
is a benign, locally aggressive tumor that accounts for
approximately 4–5% of primary bone tumors. Most
commonly, giant cell tumors are found in long bones.
CT and MRI may reveal an associated soft-tissue com-
ponent, and contrast administration may help to differ-
entiate the bony and soft-tissue elements. On MRI, they
appear as low to intermediate signal intensity on T1- and
T2-weighted imaging. Giant cell tumors often cause the
destruction of sacral foramina. Osteochondroma is an
uncommon tumor, accounting for about 4% of solitary
spinal tumors. Osteoschondromas seem to demonstrate
distinct predilection for certain parts of the spine, specif-
ically C2. It is often difficult to discover the lesion on
radiograph or MRI images due to its small size and
the continuity with the bone it originates from. Thin-slice
CT scans may be needed to confirm the diagnosis (Vidal
and Murphey, 2007).
Malignant primary spine tumors
Chondrosarcoma is the second most common nonlym-
phoproliferative malignant tumor of the spine in adults
(Rodallec et al., 2008). The mineralized matrix is best
visualized by CT, and areas of calcification are repre-
sented as signal voids on MRI. The noncalcified tumor
shows low attenuation on CT, low to intermediate signal

Fig. 33.5. Aneurysmal bone cyst (ABC). Sagittal and axial T2-weighted images (A and B respectively) show a well-defined, mul-
tilobulated lesion in the L3 vertebral body. The lesion consists of several small cystic compartments which demonstrate sharply
separated areas of T2 hyperintense and intermediate signal, representing fluid–fluid levels due to gravitational forces. This is best
demonstrated on the axial T2 scan due to the supine position of the patient (white open arrow, B). The lesion extends to the spinal
canal and causes the compression of the dural sac (white arrows). Computed tomography (CT) scan (C) shows an almost complete
replacement of osseous tissue by the lesion with only a thin layer of cortical bone remaining on the right side (black open arrow). It
also depicts the ballooning character of the lesion and some sclerotic changes on the margins. The invasive nature of the ABC is
also highlighted by the partial destruction of the upper endplate as displayed by the sagittal T2 scan (asterisk). On sagittal recon-
struction of the CT scan (D) it is also clear that the lesion extends posteriorly into the pedicle (black arrow).
Another random document with
no related content on Scribd:
adult Hymenopterous Insect it would appear that the first abdominal
ganglion is always joined with the last thoracic.

Sub-Orders.—The distinction in the form of the abdominal

articulation, previously alluded to (p. 492, Fig. 336, A, B), divides the
Hymenoptera into two great sub-Orders, the members of which are
very different in their habits and life-histories. The Sessiliventres are
plant-eaters; their larvae (Fig. 343, A) are provided with legs, and are
able to procure their vegetable food for themselves. The larvae of
the Petiolata are maggot-like and helpless, and are dependent for
food on supplies afforded them by their parents or companions. It is
said by Dewitz that although the larvae of the Petiolata appear to be
legless, there are thoracic legs within the body. The metamorphosis,
so far as it is known, and the early life-history of the Sessiliventres
are very similar to those of butterflies and moths, except that the
pupa is soft and has no hard external skin. A few of these plant-
eating Sessiliventres become carnivorous in the perfect state—a
change of habit that is most unusual in Insects, though the reverse
occurrence is common. The larvae of the Petiolata exhibit, in the
cases that have been examined, the peculiarity that the alimentary
canal has not any outlet posteriorly until the termination of the larval
stage of existence is approaching. In some cases there is no anal
orifice; in others this orifice exists, but there is no communication
between the stomach and the posterior intestine.

Packard informs us[415] that in Bombus the larva, after it is full fed,
passes into the pupa state (Fig. 331, A, B) by a series of
transformations accompanied by moultings of the skin. Packard's
statements have been confirmed by others, but details have not
been fully given, so that the number of the moults, their intervals and
other particulars, are still unknown. We have remarked that the pupal
instar is very like the perfect instar, except that it is colourless and
soft, and that each of the members is wrapped in a very delicate
skin; the colour appears gradually. This metamorphosis exhibits
important differences from that of the Lepidoptera. Packard calls the
Insect, during the stages of transformation from the full-fed larva to
the pupa, the semi-pupa; the later stages of the pupa, when the
colouring has appeared, he terms the subimago. Altogether he
considers there is a series of at least ten moultings of the skin. His
ideas were apparently derived from examination of a series of
specimens after death rather than from observation of the
development in living individuals. The parasitic forms of
Hymenoptera have apparently extraordinary metamorphoses of very
varied kinds.

Parthenogenesis.—One of the most remarkable facts connected

with this Order is the prevalence of parthenogenesis in a
considerable number of widely separated species. In many of these
Hymenoptera it is not a mere occasional occurrence, but plays an
important part in the continuity of the species; indeed, it is believed
that in some members of the Order the reproduction is entirely
parthenogenetic. We shall give particulars as to some of these cases
in subsequent chapters, and will here make some remarks on the
different forms of parthenogenesis existing in the Order. The three
forms of parthenogenesis mentioned on p. 141 all occur in
Hymenoptera. In the gall-making Cynipidae parthenogenesis is
frequently accompanied with alternation of generations, a generation
consisting of the two sexes being followed by another consisting
entirely of females, which in its turn gives origin to a bisexual
generation. In this case deuterotokous parthenogenesis is
established as a part of the normal economy of the species. This
same form of parthenogenesis also occurs in other species of
Cynipidae unaccompanied by alternation of generations. Thus in
Rhodites rosae the generations resemble one another, and the male
is very rare, but is still occasionally produced,[416] and the same
condition exists in other Cynipidae. According to the observations of
Adler, we may assume that the male, in the latter cases, is useless,
the continuation of the species being effected by virgin females
although males exist. Deuterotokous parthenogenesis also occurs in
the sawflies, but as a comparatively rare phenomenon.[417]
Thelyotokous parthenogenesis is common in sawflies, and it also
occurs in some Cynipidae. There are several species of this latter
family in which no males have ever been found.[418] The phenomena
in Rhodites rosae we have mentioned, give rise to the idea that in
that species deuterotokous parthenogenesis occurs as an exception,
the species being usually thelyotokous. A most remarkable case of
thelyotokous parthenogenesis is said to exist in the case of the
parasitic ant Tomognathus. This species is said to be monomorphic,
only the female existing, and reproducing by uninterrupted
parthenogenesis.

Arrhenotokous parthenogenesis—i.e. parthenogenesis in which the

progeny is entirely of the male sex—occurs in several species of
sawflies. We find it also occurring in the case of the social
Hymenoptera; the workers of ants, bees, and wasps occasionally
produce eggs parthenogenetically, and the progeny in these cases is
always of the male sex. In the honey-bee the queen sometimes
produces eggs before she has been fertilised, and the
parthenogenetic young are then always of the male sex.

Some species of Hymenoptera exhibit two forms of parthenogenesis.

In Nematus curtispina the parthenogenetic generation is generally of
the male sex, but a female is occasionally produced;[419] while in
Hemichroa rufa parthenogenesis may result in either deuterotokous
or thelyotokous progeny. No case is yet known of a species
exhibiting the three forms of parthenogenesis. From this review we
may conclude that parthenogenesis does not favour the formation of
one sex more than another; but it is clear that it decidedly favours
the production of a brood that is entirely of one sex, but which sex
that is differs according to other circumstances.

Production of Sex.—It is believed that a very peculiar form of

parthenogenesis exists in the honey-bee, and it is confidently stated
that the drones, or males, of that species are always produced from
unfertilised eggs. These views are commonly called the Dzierzon
theory, and are widely accepted. They assume that the eggs are
male till fertilised, and then become female. After the queen-bee is
fertilised most of the spermatozoa soon find their way into a small
chamber, the spermatheca, near the posterior orifice of the body; it is
believed that each egg may be fertilised as it passes the door of this
chamber, and that the eggs that produce females (i.e. workers or
queens) are so fertilised, but that the eggs that produce drones are
not fertilised. Hence it is supposed that the sex is determined by this
act of fertilisation, and Cheshire has described what he calls an
apparatus for differentiating the sexes. It is also confidently stated
that no male honey-bee ever has a father.

The facts we have stated as to the sexes resulting from

parthenogenetic reproduction in Hymenoptera generally, are
extremely opposed to the Dzierzon theory, in so far as this relates to
the production of sex. There have always been entomologists[420]
who have considered this view unsatisfactory, and the observations
of several recent French naturalists[421] are unfavourable to the idea
that the sex of an egg is determined by its fertilisation.

There can be no doubt that the queen honey-bee frequently

produces males parthenogenetically, and the error of the views we
are alluding to consists in taking the parthenogenesis to be the
cause of the sex of the individual. It must be recollected that the
laying of an unfertilised egg by a fertilised female may be different
physiologically from the laying of an egg by an unfertilised female;
for, though both have as result an unfertilised egg, it is possible that
the fertilisation of the female may initiate processes that modify the
sex of the eggs produced by the ovaries, so that though these may
produce previous to fertilisation only male eggs, yet after fertilisation
they may produce eggs of the opposite sex or of both sexes. In other
words, the act of fertilisation may initiate a different condition of
nutrition of the ovaries, and this may determine the sex of the eggs
produced.

Polymorphism, or Castes.—The question of the causes of the

modified individuals forming the various castes of the social
Hymenoptera has been much discussed. These individuals are
many of them very different in size and structure from either of their
parents, and are also different in their habits and instincts. This
difficult subject is far from being completely elucidated. In the case of
the honey-bee it is well established that an egg of the female sex
can, after deposition, be made either into a queen or a worker-bee
by the mode of nutrition—using that word in the largest sense. On
the other hand, Dewitz thought that in the case of the ant Formica
rufa, the caste—whether worker or winged female—is already
determined in the Insect before leaving the egg.[422] Weismann and
others associate the caste with some hypothetic rudiments they
consider to exist at the very earliest stage of the embryonic, or
oogenetic process.

Herbert Spencer says:[423] "Among these social Insects the sex is

determined by degree of nutrition while the egg is being formed,"
and "after an egg, predetermined as a female, has been laid, the
character of the produced Insect as a perfect female or imperfect
female is determined by the nutrition of the larva. That is, one set of
differences in structure and instincts is determined by nutrition before
the egg is laid, and a further set of differences in structures and
instincts is determined by nutrition after the egg is laid."

Spencer's generalisation is not inconsistent with the facts hitherto

brought to light, though it is possible that the progress of knowledge
may show some variety as to the periods of the development at
which the commencements of the modifications occur.

Fig. 339 represents the chief castes, or adult forms, existing in a

community of one of the most highly developed of the species of
social Hymenoptera, the leaf-cutting ant, Atta cephalotes. We shall,
when dealing with Formicidae, enter into some details as to these
and other cases of polymorphism.
 Fig. 339.—Adult forms of Atta (Oecodoma) cephalotes, taken from a
nest in Trinidad by Mr. J. H. Hart, 25th June 1895. A, male; B,
winged female; C-F, various forms unwinged; C, so-called soldier;
D, large worker; E, smaller worker; F, smallest worker or nurse.
All equally magnified (one and half times).

Our object at present is to bring to the eye of the reader the great
diversity of outer form that is believed, rightly or wrongly, to result
from the mode of treatment of the young. And we will also take this
opportunity of more fully illustrating the remark we made on p. 85 as
to the profound distinctions that exist between ants and white ants,
or Termites, notwithstanding the remarkable analogies that we shall
find to exist in many of their social arrangements.

The analogies we allude to, coupled with the fact that there is a
certain general resemblance in outer form between the workers of
Termites and ants, and even between the extraordinary castes called
soldiers in the two groups, have given rise to the idea that there is a
zoological relationship between the social forms of Neuroptera and
Hymenoptera. The two are, however, zoologically amongst the most
different of Insects. The external skeleton in Termites is remarkable
for its imperfect development, the sclerites being small and isolated,
while the segmental differentiation of the body is low (Fig. 225, etc.),
so that there is no difficulty in counting the segments. In ants the
reverse is the case as regards both these facts, the various
segments being most unequal, so that their homologies have only
been detected after prolonged studies, while the chitinisation and
articulation of the various parts is so complete that the ant may be
described as cased in armour, fitting together so exactly that it is
difficult anywhere to introduce the point of a needle into its chinks.
The wings of the two kinds of Insects are also extremely different.
The differences between the modes of growth and development of
the two sets of Insects are as profound as the distinctions in their
anatomy. Termitidae belong to the division of Insects in which the
wings are developed outside the body; Hymenoptera to the division
in which they are developed inside the body. In Termites the growth
of the individual is slow, and the final form is reached gradually. In
the ants the growth is carried on with great rapidity, and during it the
Insect is a helpless maggot absolutely dependent on the attentions
of its seniors, while the difference in form and structure between the
ant-larva and the ant are enormous. Both anatomy and ontogeny are
profoundly different in ants and Termites. To these distinctions must
be added, as of much importance, the fact that in Hymenoptera only
the female sex is modified for the division of labour, while in Termites
both sexes undergo this change. Hence it is impossible to suppose
that the remarkable analogies that exist between the societies of
ants and those of Termites are due to any common origin. It is
probably to some similar physiological susceptibilities in the
ancestors, at an extremely remote epoch, of both groups that we
must look for an explanation of the interesting resemblances in the
social lives of ants and Termites.

The Hymenoptera are no doubt one of the largest Orders of Insects,

the species of the parasitic tribes being apparently innumerable. No
doubt 250,000 species of the Order exist, and possibly the number
may prove to be very much larger. Up to the present time 25,000 or
30,000 have been discovered. No remains of Insects of this Order, of
older age than the Lias, have been brought to light; it is indeed
doubtful whether the fossils considered to be Hymenopterous of the
period referred to are really such.

The Order, as already mentioned, consists of two very distinct sub-

Orders, viz.:—

1. Hymenoptera Sessiliventres.—Insects with the abdomen

broad at the base, its first segment not completely amalgamated
with the thorax.

2. Hymenoptera Petioliventres or Petiolata.—The abdomen

connected with what appears to be the thorax by a slender joint,
the posterior part of the apparent thorax consisting of an
abdominal segment.

Hymenoptera Sessiliventres.—This group has been variously

called Hymenoptera phytophaga, H. securifera, H. sessiliventres, H.
serrifera, H. symphyta. We prefer an old term, taken from a
character that enables us to recognise at a glance which group a
species belongs to. The division or sub-Order may be formally
defined as follows:—

Abdomen nearly continuous in outline with the thorax, the two

parts having a broad connexion instead of a small highly mobile
articulation. Anal lobe of hind wings usually of considerable size.
Trochanters ditrochous (transversely divided into two, Fig. 345).
Extremity of body of female furnished with saws or boring
instruments, usually concealed, in some cases visible in part.
Larvae with complex mouth-parts; three pairs of thoracic legs
(imperfect in Cephidae and Siricidae), and frequently with
numerous abdominal legs, which are destitute of hooks. Food
vegetable.
The Insects of this sub-Order never exhibit the highly specialised
habits and activity of the better known petiolate Hymenoptera.
Though the food in the larval stages is always vegetable, there is
considerable variety in the larvae and their habits; some feed in
galls, some in the twigs of plants, some in the hard wood of trees
and shrubs. The majority, however, live on the leaves of plants.
Those that live in wood (Fig. 342, C) resemble in appearance
Coleopterous larvae that have similar habits, and those that live on
leaves (Fig. 343, A) resemble Lepidopterous larvae that do likewise.
There are four families included in the sub-Order, viz. Cephidae,
Oryssidae, Siricidae, Tenthredinidae.

The British Sessiliventres—under the name Phytophagous

Hymenoptera—have recently been monographed by Mr. Peter
Cameron in a series of vols. published by the Ray Society.[424]
These contain many figures and many details relating to natural
history, in addition to the descriptions of genera and species.

Fam. I. Cephidae—Stem Sawflies.

Slender Insects, with weak integument; free, more or less

elongate pronotum; one spine on the front tibia. Larvae living in
the stems of plants or in the tender shoots of trees and shrubs.

The obscure little Insects composing this family have slender

antennae of peculiar form, composed of eighteen to thirty joints, two
of which are short and stout; then come several long joints, with
more or less power of movement, the terminal portion consisting of
an elongate club of many joints with little power of movement. The
pronotum is longer than is usual in the Hymenoptera, and instead of
being very closely connected with the mesonotum, it is free and
mobile, although its base overwraps the front of the mesonotum. The
median plate (i.e. the dorsal plate connecting the thorax and
abdomen) is divided to the base along the middle, the divisions
being separated by a membranous piece broader behind; the anal
lobe of the posterior wings is small but distinct. The female bears a
saw at the extremity of the body, but it is covered by two flaps; these
form a short, terminal projection. Although too much neglected, the
Cephidae are really of great interest as being of more imperfect or
primitive structure than any of the other families of Hymenoptera.
The larval history has been traced in several species. C. pygmaeus
is sometimes very injurious to corn crops on the continent of Europe,
and even in our own country its effects in this respect are considered
to be occasionally serious. The egg is laid in the stem of the corn
plant; the larva soon hatches and eats its way upwards in the stem.
It is a soft grub, apparently footless, but really possessing six small
projections in place of thoracic legs. It occupies all the summer in
feeding, and when full fed and about to prepare for its
metamorphosis, it weakens the stem by a sort of girdling process
below the ear; it then descends in the stem to near the root, where it
constructs a transparent cocoon, in which it passes the winter as a
larva, changing to a chrysalis in the month of May, and completing its
development by appearing as a perfect Insect shortly thereafter. The
girdling operation is very injurious, and causes the corn stem, when
ripe or nearly so, to break in two under the influence of a strong
wind, so that the ears fall to the ground.

Fig. 340.—Cephus pygmaeus. Upper figure, larva; lower, female

imago. Britain. (After Curtis.)

The history of C. integer has been given by Riley. This Insect attacks
the young shoots of willows in North America. Riley states[425] that
by a wonderful instinct the female, after she has consigned her egg
to the twig, girdles the latter, preventing it from growing any further,
and from crushing the egg by so doing. The larva after hatching eats
downwards, sometimes destroying a length of two feet of the twig;
when full grown it fills the bottom of the burrow with frass, and then
previous to making its cocoon eats a passage through the side of the
shoot about a quarter of an inch above the spot where the cocoon
will be placed, thus making it easy for the perfect Insect to effect its
escape; it leaves the bark, however, untouched, and is thus
protected in its retreat. A delicate transparent cocoon is then spun in
which the larva passes the winter, changing to a pupa in the
following March, and emerging as a perfect Insect about six weeks
thereafter.

Somewhat less than 100 species of this family are at present known;
the great majority are found in the Mediterranean region, but there
are several in North America. As a single species is known from
Mexico and another from Japan, it is probable that the family may
prove to have a wider geographical extension than at present
appears to be the case.

Fam. II. Oryssidae.

The median plate behind the metanotum entire, not divided in

the middle; antennae inserted below the eyes immediately
above the mandibles, under a sharp edge.

This family consists of the genus Oryssus, and includes only about
twenty species, but is nevertheless very widely distributed over the
world. They are very rare Insects, and little is known as to their
habits; one species, O. abietinus, was formerly found in England.
Should any one be so fortunate as to meet with it, he can scarcely
fail to recognise it on noticing the peculiar situation of the base of the
antennae. In this respect the Chrysididae somewhat resemble
Oryssus, but in that group of Hymenoptera the hind body or
abdomen is remarkably mobile, so that the Insects can coil
themselves up by bending at this joint; whereas in Oryssus the hind
body is very closely amalgamated with the thorax—more so, in fact,
than in any other Hymenopterous Insect—and has no power of
independent movement.

Fig. 341.—Oryssus sayi. North America. A, The female Insect; B, head

seen from the front.

Oryssus abietinus very closely resembles C. sayi (Fig. 341); it has

indeed been recently suggested by Mr. Harrington that the two
supposed species may really be identical.

Fam. III. Siricidae or Uroceridae.

Pronotum closely connected with the mesonotum, perpendicular

in front; the anterior lobe of the latter not separated by the lateral
lobes from the posterior lobe: the median plate (behind the
metathorax) is divided longitudinally along the middle. The
female is provided at the extremity of the body with an elongate,
cylindrical boring instrument. The larvae live in the wood of
trees.
 Fig. 342.—Tremex columba. North America. A, Imago, female: B,
pupa, female, ventral aspect: C, larva; a, imperfect legs: D,
parasitic larva of Thalessa. (B and D after Riley.)

The Insects of this family are usually of large size and of bright
conspicuous colours; these, however, frequently differ greatly in the
sexes of the same species, and may be very variable even in one
sex. The antennae are filiform and usually elongate; the head is
usually contiguous with the thorax, but in one division, Xyphidriides,
it is exserted and separated from the thorax by a well-marked neck.
The pronotum is attached to the mesonotum, and possesses very
little, if any, freedom of movement; it varies in its size, being
sometimes conspicuous from above; in the Xyphidriides it is smaller,
and in the middle is entirely vertical in its direction. The mesonotum
is moderate in size, and its divisions are delimited by broad vague
depressions. The prosternum appears to be entirely membranous,
but the prosternal plates (pleura) are large, and meet together
accurately in the middle, so as to protect the greater part of the
under-surface of the neck. The abdomen is cylindrical or somewhat
flattened above; it has seven dorsal plates in addition to the spine-
bearing terminal segment. The trochanters are double, the outer
division being, however, short; the anterior tibia has only one spur;
the anal lobe of the posterior wings is large. The "borer" or ovipositor
of the female is a remarkable organ; it is held projecting directly
backwards from the extremity of the body, and has the appearance
of being a powerful sting. The apparatus is much longer than it
appears, for it proceeds not from the apex of the body, but from the
under-surface far forwards, so that the part exposed is only about
one-half of the total length; it consists of a pair of elongate sheaths,
which are easily separable though they wrap together, and enclose a
slender tube. This tube is rigid and quite straight; though appearing
solid, it is really composed of two very perfectly adjusted laminae
and a third arched piece or roof. The two lower laminae are called
the spiculae; they are serrated or grooved in a peculiar manner near
the tip, and although so closely adjusted to the borer or upper piece
of the tube as to appear to form one solid whole with it, they are said
to be capable of separate motion. In addition to these parts, the
termination of the abdomen bears above a shorter piece that
projects in a parallel plane, and forms a sort of thick spine above the
ventral pieces we have described; this process is very strong, and
has in the middle of its under-face in Sirex gigas a membranous
cavity, replaced in S. juvencus, according to Westwood, by a pair of
minute pilose styles. The Insect, by means of this powerful
apparatus, is enabled to deposit her eggs in the solid wood of trees,
in which the larva sometimes penetrates to the depth of eight inches.

Sirex gigas is one of the most remarkable of our British Insects, but
is little known except to entomologists, being usually rare. On the
continent of Europe it is, however, an abundant Insect, especially in
the neighbourhood of forests of fir-trees, and is a cause of
considerable terror. As the Insect is not capable of inflicting much
injury to the person, it is probable that the peculiar ovipositor is
believed to be a sting. The eggs are laid—it is said to the number of
100—in the solid wood of fir-trees, but not in perfectly healthy wood;
the reason for this, it is thought, being that in a healthy tree the great
affluence of sap caused by the burrows and presence of the Insect
would be injurious to the latter. The Sirex will, however, attack a
perfectly healthy tree immediately after it has been felled. The larva,
small at first, enlarges its burrows as itself grows larger, and thus the
wood of a tree may be rendered completely useless for trade
purposes, although there may be very little outward indication of
unsoundness. The larva (Fig. 342, C, larva of Tremex) is a pallid,
maggot-like creature, with six projections representing thoracic legs;
there are no other legs behind these, but some slight protuberances
take their place; the terminal segment is enlarged, and bears a hard
spine. There is a difference of opinion as to the duration of the life of
the larva, Kollar saying that in seven weeks after the deposition of
the egg the maggot is full fed, while others consider that it takes two
years to attain this condition; the latter statement is more probably
correct, it being the rule that the life of wood-feeding larvae is more
than usually prolonged. After becoming full fed, the Insect may still
pass a prolonged period in the wood before emerging as a perfect
Insect. As a result of this it not infrequently happens that the Insect
emerges from wood that has been carried to a distance, and used
for buildings or for furniture. A case is recorded in which large
numbers of a species of Sirex emerged in a house in this country
some years after it was built, to the great terror of the inhabitants.
The wood in this case was supposed to have been brought from
Canada.

Fabre has studied[426] the habits of the larva of Sirex augur, and
finds that it forms tortuous galleries in the direction of the longitudinal
axis of the tree or limb, and undergoes its metamorphosis in the
interior, leaving to the perfect Insect the task of finding its way out;
this the creature does, not by retracing its path along the gallery
formed by the larva, but by driving a fresh one at right angles to the
previous course, thus selecting the shortest way to freedom. By what
perception or sense it selects the road to the exterior is quite
unknown. Fabre is not able to suggest any sort of perception that
might enable the larva to pursue the right course, and considers it
must be accomplished by means of some sensibility we do not
possess. Fabre's observation is the opposite of what has been
recorded in the case of S. gigas, where the larva is said to prepare
the way for the exit of the perfect Insect.

Individuals of Sirex are often found in dried and solid wood, encased
by metal. When the Insect finds itself so confined, it gnaws its way
through the metal, if this be lead, and escapes. The perseverance
displayed by the Insect in these circumstances seems to indicate a
knowledge of the direction in which liberty is to be found.
About 100 species of Siricidae are known. They form two sub-
families:—

1. Siricides: back of head nearly or quite contiguous with the

pronotum.

2. Xyphidriides: back of head separated from the pronotum by

an elongate neck.

We are reputed to possess in Britain two species of each of these

sub-families, but it is doubtful whether more than one Siricid is truly
native. Sirex gigas is frequently brought over in timber, and certainly
breeds at times freely in Britain. Mr. Leech has recorded the
occurrence of the larvae in abundance in fir-trees in the
neighbourhood of Dublin. Sirex juvencus is more rarely met with.
Xyphidria camelus is doubtless a native, though now apparently rare.
It used to occur about old willows, near London, in the New Forest,
and, I believe, also in the neighbourhood of Cambridge.

Fam. IV. Tenthredinidae—Sawflies.

Hymenoptera Sessiliventres, having the pronotum small,

accurately adapted to the mesonotum; the anterior lobe of the
latter is widely separated from the posterior; there are two spurs
on the anterior tibiae. The larvae usually live on leaves after the
manner of caterpillars, but a few inhabit galls.

The sawflies are an important family of Insects, their species being

numerous, while some of them are, in the larval state, very
destructive to vegetables and fruit. Being quiet creatures, rarely seen
on the wing, they are, though common Insects in this country, but
little known, and few persons recognise a sawfly as such. They are
usually of small or moderate size, and the numerous species have a
great family resemblance. This remark requires some qualification in
the case of the Cimbicides, they being Insects of larger size—usually
surpassing the honey-bee—of more robust structure, and with
greater powers of flight.

Fig. 343.—Lophyrus pini. Britain. A, Larva; B, ventral aspect of pupa;

C, imago, male. (After Vollenhoven.)

The antennae are remarkably variable in form and structure.

Cameron considers that nine should be taken as the normal number
of their joints; but there are only three in Hylotoma, while in Lyda
there may be forty or more. The head is usually held closely applied
to the thorax, but is really borne on a neck capable of much
elongation (Fig. 332). The pronotum forms a part of the alitrunk, but
is not soldered thereto. Usually the prosternum is more or less
completely concealed by the side-pieces, but in Cimbicides it is
larger and conspicuous, the side-pieces being in this group smaller
than usual. The dorsal pieces of the mesothorax have their relative
proportions different to what we find them in the other families of
Sessiliventres, and even in most of the other Hymenoptera. There is
first an antero-median lobe of triangular shape projecting, like a
wedge, far backwards, into the great lateral lobes. These latter form
the larger part of the area of the mesonotum; they meet together in
the middle line, and behind are separated by a deep depression from
the posterior lobe, or scutellum of the mesothorax, which is
frequently divided into two parts, the anterior being the so-called
scutum. The pieces of the metanotum are short and obscure, owing
to the great unevenness of their parts; on each side of the middle
there is a small membranous space of pallid colour. The cenchri, as
these spaces are called, are, in Lyda, delicate, membranous,
depressed spaces, in front of each of which there stands up a flap of
membrane. The function of the cenchri is quite unknown. The
median plate is fastened to the hind margin of the metanotum, and
looks quite like one of the dorsal plates of the following abdominal
segments, from which, however, it is separated by a more or less
conspicuous membrane. In the majority of the Tenthredinidae the
median plate is divided along the middle, but in the Cimbicides this is
not the case. The mesosternum is very large, and the metasternum
small, so that the middle and hinder pairs of coxae are placed close
together. The abdomen consists of nine segments, there being eight
dorsal plates in addition to the median plate, and seven ventral
plates besides the terminal armature. There is a pair of short cerci,
each of a single segment. The trochanters are divided; each tibia
bears two spurs at the extremity, and the tarsi are 5-jointed.

Fig. 344.—Saws of Cimbex sylvarum. A, The pair spread out and

placed in a horizontal position; a, the lower margin of the saw
proper; b, the upper margin of the support: B, two teeth of the saw
more highly magnified.

The most characteristic and interesting of the structures with which

the Insects of this family are provided is the apparatus from which
the name of sawfly is derived. As long as two centuries ago these
instruments excited the admiration of Vallisnieri and of Réaumur,
who described them at length; and it is truly astonishing that any part
of a living being should be changed into tools so mechanically
perfect as these saws are (Fig. 344). They serve the purpose of
assisting the female in depositing the eggs in a suitable situation, the
place selected being frequently the tender stems of shrubs or other
plants, or the interior of leaves. These organs are therefore of course
possessed only by the female. They are placed on the lower aspect
of the hinder extremity of the body, where they are enclosed and
protected by a pair of sheaths, from which they can be made to
protrude by a little pressure exercised on the parts immediately in
front of them. Each female possesses a pair of these saws; they
consist of thin laminae of very hard consistence, and are not only
toothed at their edge, but in many cases each tooth is itself serrate;
at the same time the outer face of the saw is sculptured or plicate in
a remarkable manner, so that the saw in this way acts as a file or
rasp. The Insect having selected a suitable place, uses the saws by
placing the extremity of the abdomen against a twig or leaf,
protruding the blades, which, moving with an alternate motion, one
being thrust forward while the other is retracted, act on the plant so
as to make an incision. Each saw is directed in its movement by the
support, the pair of supports being united at the base by membrane
as shown in Fig. 344. In the case of some species,—Hylotoma
rosae, the common sawfly of our rose-bushes, for instance—there is
no difficulty in observing the operation; indeed old Réaumur, when
speaking of the placid disposition of the sawflies, suggests that it
was given them so that we may easily observe their charming
operations. We cannot but regret that in these days we are unable to
take so complacent a view of the arrangements of nature. There is
much variety in the details of the structure of these saws; so much
indeed that it is possible to identify most of the species by means of
the saw alone. According to certain observers, the eggs are laid by
some kinds on, not in, the leaves, so that we may conclude that in
these cases the saws are not used by their possessors. An incision
having been made, an egg is placed in it, and also a drop of some
liquid matter. The egg is at first small, but soon increases till it
becomes twice or three times its former size, and the development of
the embryo commences.

The larvae of the Tenthredinidae exhibit great variety, and are indeed
in this respect more interesting than the perfect Insects. The usual
rule is that the larvae much resembles those of Lepidopterous
Insects, and feed exposed on plants in the same way as
Lepidopterous larvae do. But the exceptions are numerous;
sometimes the larva is covered with slime, and thus protected from
various enemies. In other cases it is very depressed, a broad
creature, of irregular outline, living closely attached to the leaf,
somewhat after the fashion of a huge scale-Insect. Some larvae
mine between the layers of a leaf, others roll up leaves; a few live in
the stems of plants, and one or two inside fruits. Even this does not
complete the list of their habits, for a few species of Nematus live in
galls caused by the deposition of the egg. A species of Lyda forms
for itself a case out of bits of leaves, and carries this habitation about
with it after the fashion of the Phryganeidae. The number of legs in
these larvae is unusually great, varying from eighteen to twenty-two
—that is, three pairs of thoracic legs and eight of abdominal or pro-
legs. This character offers a ready means of distinguishing, in the
majority of cases, these larvae from those of the Lepidoptera in
which the number of legs varies, but is only from ten to sixteen;
moreover, the pro-legs in sawflies are destitute of the circles of
hooklets that exist in Lepidoptera. This mode of identifying the
immature stages of the Tenthredinidae is not, however, always
satisfactory, as there are some of these larvae that have no pro-legs
at all, but only the three thoracic pairs. Another point of distinction
exists, inasmuch as the larvae of the sawflies have only one ocellus
on each side of the head, whereas in the Lepidopterous caterpillars
the rule is that there are several of these little eyes on each side. In
addition to this, we should mention that the Lepidopterous larva
never has any pro-legs on the fifth body-segment, whereas in the
sawflies when pro-legs are present there is always a pair on the
segment in question.

These larvae are of various colours, but the patterns and markings
they exhibit are not quite like those of the Lepidoptera, though it
would be difficult to make any correct general statement as to the
nature of the differences. The variety of their postures is very
remarkable; and in respect of these also Tenthredinidae differ
considerably from Lepidoptera. Some of them hold the posterior part
of the body erect, clasping the leaf by their anterior legs; others keep
the posterior part of the body curled up (Fig. 343, A), and some
combine these methods by curving the posterior part of the body and
holding it away from the food. These attitudes, like the general form,