Potential of lumbodorsal fascia forces to generate back extension moments
during squat lifts
Dear Sir,
The main conclusion of McGill and Norman’-that
what we once termed the active part of the
lumbodorsal fascia or the hydraulic amplifier
mechanism’, does not provide a substantial extension
moment in the sagittal plane-is correct, but
somewhat out of date3s4. Nevertheless the paper
deserves some comment, as its authors have made a
separate and highly contentious claim concerning
the role of the ligaments in sagittal plane lifting,
without adequate support. Due to the potential
influence of this model (for which the authors won
the Volvo prize in 1986) on clinical practice and
biomechanical thought, we have undertaken
related correspondence with Spine, where details were
first published.
The authors claim that because none of their
subjects exhibited a total lumbar flexion of more than
32” (of which they would ascribe 29% or 9.3” to
L45), the seven passive ligaments which they included
in their model could not be set under tension, and
thus could not contribute a sagittal-plane extension
moment. It must be emphasized that this is the critical
assumption/calculation upon which all subsequent
calculations of their model depend, for the authors
assign all the leftover moment (which under their
assumptions on ligament elongation, turns out to be
all the moment) to the activity of the muscles. The
EMG readings are only used to assign what
proportion of the load should go to which muscle,
not to determine whether or not the muscles are in
fact supporting the entire load. Therefore it should
not come as a surprise that with only muscles in the
final set of assumptions, only muscles are found in
the solution.
Thus it is this calculation that lumbar flexion
is inadequate to set the connective tissue under
tension that must be examined more closely. First of
all, it conflicts with much experimental data on the
response of the motion segment. Consider McGill
and Norman’s figure of 9.3’ of flexion for L4. Adams
et al6 found that the 6 L4-L5 motion segments they
studied had flexion limits of between 7 and 12’ (mean
10 + 1.8” SD) ; and that the ligaments and disc of
these samples resisted between 22 and 88 Nm of
bending moment (mean = 46 f 24 Nm SD). Any
contribution by passive stretching of the other
connective tissues (lumbodorsal fascia, skin, etc.)
would be in addition to this. Thus not only were the
connective tissues of the motion segment experi-
mentally determined to be supporting a significant
moment, but the degree of lumbar flexion which
McGill and Norman claim does not set the ligaments
under tension turns out to have been close to or at
the physiologic limits.
0 1989 Butterworth & Co (Publishers) Ltd
0141-5425/89/020172XI5 $03.00
172 J. Biomed. Eng. 1989, Vol. 11, March
What is the reason for this discrepancy? The
answer to this question cannot be found in ref. 1,
nor in McGill and Norman’s ( 1986) Volvo Prize
article5. It can be obtained, however, by rooting
about in McGill’s (1986) PhD thesis7.
According to McGill’ (p. 92), ‘Ligaments con-
straining flexion are slack during the beginning of
lumbar flexion. The degree of flexion necessary for
the ligaments to start producing force from strain is
critical for the modeller.’ It seems that McGill’
decided that this degree of flexion should be 14”.
The basis for this choice was that the constitutive
a equations used by McGill to predict force from strain
would thereby limit the maximum flexion of the
L4/L5 joint to about 17 “, which was taken by McGill
to be a typical value for the maximum flexion at this
level. This estimate was in turn arrived at by
assuming the typical maximum flexion of the lumbar
spine to be 60”, of which 29 % would come from L4.
The figure of 60” was attributed (McGill’, p.148) to
Farfans. Thus with McGill’s subjects never exhibiting
more than 32” of total lumbar flexion, or 9.3” at
L4/L5, the neutral value of 14” would never be
exceeded, and the ligaments would never experience
any elongation past their resting state.
But 60” is a huge figure for maximum lumbar
flexion. Macrae and Wright9 (Figures 5 and S), using
X-ray techniques, reported no values greater than
40”. Ifone applies their regression model (correlation
coefficient = 0.97, SD = 3.5”) for measuring lumbar
flexibility from the modified Schober test to the data
of Batti’e et ~1.” on 3020 subjects, a similar mean
value (35’ in ZO-29-year-old males) for maximum
lumbar flexion is obtained. Batti’e et d’s 95th
percentile ZO-29-year-old male had a maximum
lumbar flexibility of 46” according to this relation-
ship. As stated above, Adams et aL6 found an average
flexibility of 10 + 1.8” at the L4/L5 joint. If one
returns to Farfar?, one finds in table 3.1 (p. 46)
values for the ttypical range of motion of lumbar
intervertebral joints’. The value for L4/L5 is listed
at 12” in flexion, and the value for the entire lumbar
spine totals to 39.5”. The value of 60” does not seem
to be in our edition.
Thus McGill and Norman’s elimination of the
ligamentous contribution is based entirely on the idea
that total ligament slack existed at up to 14” offlexion
at L4/5. McGill7 found that if this slack ended instead
at IO”, then the ligaments of the intervertebral joint
would indeed make a substantial contribution to the
restorative moment. This is essentially the conclusion
that we reached in 19772. Given the fact that the
14” figure would seem to represent the extreme of
human variability, and the fact that its source is
 Letters to the Edifor

unknown, it would seem reasonable that McGill and fosters creativity and generates a vigorous and
Norman re-evaluate this assumption. healthy debate. However, McGill and Norman have
Furthermore, the lumbodorsal fascia, which concentrated the bulk of their article on a subject
McGill and Norman acknowledge to be connected which is no longer controversial. This only ends up
to the ilium and the spinous processes, does not diverting attention from the crucial claim-based on
seemed to have been included as a restraining element an erroneous assumption about the flexibility of the
that could be set under tension by passive stretching. lumbar spine-that postural-dependent stretching of
No reason was given for this exclusion; yet the size the connective tissues (ligaments, fascia, et al.) has
and strength of this tissue, coupled with its relatively no role to play in sagittal-plane stabilization of the
long level arm, makes this a rather serious omission. spine.
Finally, the constitutive equations of the ligaments
used by McGill and Norman to determine the Serge Gracovetsky
response to stretching do not include any time Department of Electrical Engineering,
variables. As the ligaments are visco-elastic, their Concordia University,
resistance to stretching is higher when this stretching 1455 de Maisonneuve Blvd West,
Montreal,
occurs rapidly. This factor could be significant yet
Quebec H3G IM8
was not incorporated in the model.
Canada
In sum, the model of the spine used by McGill
and Norman conflicts with published experimental 1. McGill SM, Norman RW. Potential of lumbodorsal fascia
data on the mechanical response of the intervertebral forces to generate back extension moments during squat lifts.
joint, and on the typical range of motion of the J Biomed Eng 1988, 10, 3128.
lumbar spine. It does not include the possibility of 2. Gracovetsky S, Farfan HF, Lamy C. A mathematical model
the lumbodorsal fascia acting as a passive restraining of the lumbar spine using an optimized system to control
element. It also conflicts with the fact that one of muscles and ligaments. Orthop Clin N Am 1977,8(l), 135-53.
3. Macintosh JE, Bogduk N, Gracovetsky S. The biomechanics
the best ways to measure lumbar flexion is in fact
of the thoracolumbar fascia. Clin Biomech 1987, 2, 78-83.
the significant skin elongation seen in the Schober
4. Tesh KM, Shaw Dunn J, Evans JH. The abdominal muscles
testsg,lO. Given the extreme sensitivity of their model
and vertebral stability. Spine 1987, 12(5), 5014.
to the assumed limit oflumbar flexion, and the nature 5. McGill SM, Norman RW. Partitioning of the L4-L5
of biological variability, it represents a serious dynamic moment into disc, ligamentous, and muscular
omission that this factor received no discussion at all components during lifting. Spine 1986, 11(7), 666-78.
in either ref. 1 or ref. 5. Why did they not simply 6. Adams MA, Hutton WC, Stott JRR. The resistance to
measure the maximum lumbar flexion of their actual flexion of the lumbar intervertebral joint. Spine 1980, 5(3),
test subjects? If passive stretching of the lumbodorsal 245-53.
fascia, and more reasonable values of maximal 7. McGill SM. PhD Thesis, University of Waterloo, 1986.
8. Farfan HF. Mechanical Disorders of the Low Back. Philadelphia,
lumbar flexion, are incorporated into an otherwise
Lea and Febiger, 1973.
reasonable model, McGill and Norman’s main
9. Macrae IF, Wright V. Measurement ofback movement. Ann
conclusions about the partitioning of the L4/L5 rheum Dis 1969, 28, 584-9.
dynamic moment into muscular and ligamentous 10. Batti’e MC, Bigos SJ, Sheehy A, Wortley MD. Spinal
contributions are completely reversed. flexibility and individual factors that influence it. Physical
Controversy and criticism is welcome, as it Therapy 1987, 67(5), 653-8.

Dear Sir,

The comments stated in Dr Gracovetsky’s letter to
the editor do not address the paper published in the
Journal of Biomedical Engineering ( 1988), 10 : 3 12-8.
Readers of the Journal of Biomedical Engineering will
have to read our article in Spine’ and Dr
Gracovetsky’s and Farfan’s in Spine2 to understand
what the issues are. The critique by Stokes et aL3 of
the Gracovetsky and Farfan paper is also informative.
Thus, we are surprised that Dr Gracovetsky’s letter
is being considered for publication in the Journal of
Biomedical Engineering. However, since you kindly
gave us the opportunity to respond, we are doing so.
What has prompted Dr Gracovetsky to attack
our work here, and in other forums, is that the output
from opr model of the lumbar spine shows very little
contribution from passive tissues (e.g. ligaments,
lumbodorsal fascia) to the support of the extensor
moment of the trunk in the types of lifts that we have
reported-quat lifts. His model output predicts that
these passive tissues play a major role. The reason
for the discrepancy in outputs, a discrepancy which
concerns us as much as it does Dr Gracovetsky, is
because of a dramatic difference in modelling
approach.
The major difference between our model approach
and that of Dr Gracovetsky is that he elected to use
an optimization technique to partition loads among
the ligaments and muscles of the low back. The
output of an optimization model depends, to a large
extent, on what variable or function one assumes
that the body minimizes or optimizes. He has
assumed that stress on the joint is minimized. This
assumption demands that tissues with the largest
moment arms must produce the majority of the

J. Biomed. Eng. 1989, Vol. 11, March 173

extensor moment. These tissues, according to Dr
Gracovetsky, are the lumbodorsal fascia and the
midline ligaments. However, in our opinion, no one
knows what the optimization function really is nor
can this approach respond to variations in the way
that different individuals perform lifts. If the kine-
matic input is the same on different trials, the model
output will be the same. Our model, on the other
hand, makes no assumptions about what is optimized
or minimized. Rather it is driven, trial to trial, lift
to lift, subject to subject from measured biological
responses, such as EMG to assist with allocation of
forces to the various muscles and measured thorax
and pelvis motion to try and calculate the straining
of ligaments and discs in living human subjects. Our
model is free to choose which tissues support the
moment in any given lift, depending upon the
amount of spine flexion to recruit the passive tissues,
and amount and source of muscle electrical activity
as an indication of muscle involvement.
Nonetheless, his letter has been useful because
it has forced us to go back and reassess our
assumptions. We have done this and have re-checked
data used in our model development. None of Dr
Gracovetsky’s arguments has changed our minds.
To reiterate, the output of our model shows that for
the types of lifts reported, the ligaments played very
little role. During flexion, we measured the forward
bending rotation to be predominantly about the hip
joints and not caused by significant flexion of the
lumbar spine. Further, the presence of high levels of
extensor muscle EMG activity throughout the
duration of all lifts provided no experimental support
for Dr Gracovetsky’s notion that the ligaments
alleviate the muscles of their responsibility to support
the Aexion moment. Furthermore, to our knowledge,
he has not reported convincing experimental data
that would support such a notion.
We have addressed the specific points raised
in Dr Gracovetsky’s letter in the following list.
1. Our conclusions that the limbodorsal fascia
provides no active support to extensor moment
generation were termed ‘somewhat out of date’
by Dr Gracovetsky. He referred to two articles
acintosh, Bogduk and Gracovetsky’ and Tesh
:Kl/ 5 as studies which arrived at the same
con&ions as our paper, published in the Journal
of Biomedical Engineering in 1988. Although we may
have been a little slow in scouring the current
literature, we make an attempt to keep up to date.
Both of these papers appear to have been
published after we had submitted our paper _to
the Jourraal of Biomedical Engineering. Furthermore,
the problem of lags in publication of submitted
manuscripts is well known to all of us, including
Dr Gracovetsky. Although he retracted, or at least
clarified, his claims of the major role of the
lumbodorsal fascia mechanism in providing a
substantial extension moment, in the article
which he co-authored with Macintosh and
Bogduk4, there is no mention of this retraction or
clarification in his book published in 19886.
2. Dr Gracovetsky has chosen to attack the choice
of 14” as the approximate L4/L5 intervertebral
174 J. Biomed. Eng. 1989, Vol. I 1, March
disc angle required before ligaments are appre-
ciably recruited, an angle reported in the McGill
thesis. We were negligent in not reporting this
angle in the Spine article2 and for this oversight
we apologize. From his reading of the thesis, Dr
Gracovetsky then assumed that this angle was
applied for every subject. The wording in the
thesis was not perfectly clear on this point. In fact,
the maximum amount of lumbar flexion was
measured in each subject and the ligaments of the
model were turned to contribute to resisting
flexion at the appropriate measured angle. We did
not apply an assumed angle for every subject. It
so happened that these initial three subjects
reached a similar full flexion of f 7’. Subjects can
reach full flexion while standing by hanging on
their ligaments with no EMG activity in either
the erectors or the abdominals. The ligaments and
discs support this relatively low moment at full
flexion and they will not permit significant further
flexion due to their stiffness. In 23 young male
university student subjects that have subse-
quently been analysed in our laboratory the
maximum average Aexion at L4/L5 was 16.6” (SD
2.2). While ligaments are recruited approximately
half-way to full flexion sign&ant ligament contri-
bution does not occur until within a few degrees
of full llexion. If an inflexible subject was tested
who could only manage 8” of rotation during the
full Aexion, for example, then the model would
‘tune’ the ligaments to resist further rotation.
Thus, is it erroneous for Dr Gracovetsky to state
‘the ligamentous contribution is based entirely
on the idea that total ligament slack existed at up
to 14” of flexion at L4/L5’. Nor have we ever
stated that the ligaments ‘have no role to play
in stabilization of the spine’. Anyone who slowly
‘slumps’ in a chair can palpate the ligaments
as they strain close to full flexion. The ligaments
are paramount in providing this function.
3. Dr Gracovetsky chose to quote the work of
Adams et al.’ as stating that the flexion limit of
L4/L5 was between 7 and 12” (mean 10” +_ 1.8”
SD). Careful review of the paper shows that these
data were ofin uitro specimens set in dental plaster.
However, in the same paper, Adams et al. reported
more relevant data: the average flexion limit of
the L4/L5 joint in 18 normal subjects (ages 20-58
years) was 14” +_ 3” as determined from roent-
genograms. .Further, the flexion moment applied
to these joints ranged from only 22 to 88 Nm which
could hardly be considered to support the majority
of the load when we routinely observe 450 Nm
during demanding lifts, in the laboratory. Why
would Dr Gracovetsky choose to substantiate his
point on cadaveric data when data on live
humans appeared earlier in the article. Adams
and Hutton* in a more recent paper clearly show
the difference in range of motion at the L4/L5
and other discs between cadavers and living
people.
4. The statement of Dr Gracovetsky that the passive
contributions of the lumbodorsal fascia have been
ignored in our work and ‘makes this a rather
serious omission’ is false. This passive component
 ,?_ttiersto the Editor

of the LDF is shown in equation (6) of McGill However, within the constraints of this uncertainty
and Norman’ and explained in the text. we stand by our conclusions. Our model output shows
clearly that passive tissues play almost no role in the
We have no delusions that our measurements are types of lifts that we reported. Maybe passive tissues
error-free and that all of our modelling assumptions will prove to be important in other types of lifts.
will withstand the tests ofnew experimental evidence.
As soon as our errors are proven to us we will change SM. McGill
the appropriate details in the model and our entire R.W. Norman
approach, if necessary. Dr Gracovtsky certainly has Faculty of Human Kinetics and Leisure Studies
Department of Kinesiology
not convinced us. Indeed we are beginning to resent
University of Waterloo
his selective presentation and extrapolations from
Waterloo
other people’s data and misquoting of our work to Ontario N2L 3Gl
try to substantiate his a priori position, implicit in his Canada
optimization approach, that passive tissues play a
major role in supporting the trunk during all lifting
1. McGill SM and Norman RW. Partitioning of the L4/L5
efforts.
dynamic moment into disc, ligamentous and muscular
Dr Gracovetsky continues to under-estimate the components during lifting. Spine 1986; 11, 66678.
importance of the spine musculature in main- 2. Gracovetsky S and Farfan H. The optimum spine. Spine 11,
taining dynamic equilibrium taking the viewpoint 1986; 543-73.
that the muscles are not of sufficient size. His opinion 3. Stokes IAF, Krag MH, Wilder DG. A critique of ‘the
is base on measures of muscle cross-sectional area optimum spine’. Spine 1987; 12, 51 l-2.
obtained from cadavers. This is in spite of several 4. Macintosh JE, Bogduk N, Gracovetsky S. The biomechanics
recent papers reporting the much larger muscles of the thoracolumbar fascia. Clin Biomech 1987; 2, 7883.
measured in living, working individuals. 5. Tesh KM, Shaw Dunn J, Evans JH. The abdominal mucles
and vertebral stability. Spine1987; 12, 5018.
We are confident that we have made no glaring
6. Gracovetsky S. TheSpinal Engine, New York: Springer-Verlag,
errors in the development of the model and that the
1988.
general approach is sound. Unfortunately it is not 7. Adams MA, Hutton WC, Stott JRR. The resistance to flexion
yet possible to directly validate our model outputs, of the lumbar spine. Spine 5, 245-53.
Dr Gracovetsky’s model outputs or anybody else’s 8. Adams MA and Hutton WC. Has the limbar spine a margin of
at this point in the development of the science. safety in forward bending? Clin Biomech 1986; 1, 34.

Circumferential and longitudinal viscoelasticity of human iliac arterial

segments in vitro

Dear Sir,

First let me express my warm congratulations to the pressure of a magnitude equal to the mechanical
authors of ‘Circumferential and longitudinal visco- pressure acting in vivo. Observations similar to this
elasticity of human iliac arterial segments in vitro’ or to that expressed by Papageorgiu and Jones should
(G.L. Papageorgiu and N.B. Jones, J Biomed Eng be of fundamental importance to researchers when
1988; 10: 82-90) for their excellent and very attempting to extrapolate results from in mortem
thorough treatment of this important biomechanical samples to the in vivo system. It would also be
problem. interesting to investigate the possibility of occurrence
I agree wholeheartedly with the authors’ statement of ‘creep’ and ‘relaxation’ phenomena in arteries.
to the effect that ‘there is an interesting question with When analysing static circumferential and longi-
regard to the effect of wall smooth muscle activity tudinal elasticity the authors present their data using
on wall elasticity; when cadaveric arteries are used Lagrangian definitions of strain, i.e.
there is no smooth muscle activity’. On the other
hand I also feel that the complex fluid mechanics-
thermodynamic characteristics of the blood flow may
play an important role in the mechanical properties
of the wall arteries.
It may be of some interest to point out that
Kobayashi and his colleagues’ showed that consider- where lr is the firal length of the specimen and Zi is
able viscoelastic effects were observed in cadaveric the initial length of the specimen.
cornea but that these characteristics almost disap- One might question whether it would not have
peared if the in mortemeye was subjected to an internal been more convenient from a practical viewpoint,

J. Biomed. Eng. 1989, Vol. 11, March 175