Download as pdf or txt
Download as pdf or txt
You are on page 1of 12

The Astronomical Journal, 151:103 (12pp), 2016 April doi:10.

3847/0004-6256/151/4/103
© 2016. The American Astronomical Society. All rights reserved.

EARTH’S ROTATIONAL DECELERATION: DETERMINATION OF TIDAL FRICTION INDEPENDENT OF


TIMESCALES
Steven D. Deines1 and Carol A. Williams2,3
1
Donatech Corporation, Fairfield, IA 52556, USA; steven.deines@gmail.com
2
Department of Mathematics and Statistics (Prof. emeritus), University of South Florida, 4202 E. Fowler Ave., Tampa, FL 33620, USA; cw@math.usf.edu
3
Retired Adjunct Professor of Astronomy, University of Florida, Gainesville, FL 32611, USA
Received 2013 December 5; accepted 2015 November 9; published 2016 March 29

ABSTRACT
This paper determines Earthʼs rotational deceleration without relying on atomic or ephemeris timescales. Earthʼs
rotation defines the civil time standard called Universal Time (UT). Our previous paper did not examine tidal
friction indepth when analyzing the timescale divergence between UT and International Atomic Time (TAI). We
examine all available paleontological fossils and deposits for the direct measurements of Earthʼs past rotation rates,
because that record includes all contributing effects. We examine paleontological reports that date Earthʼs rotation
rate using corals, bivalves, brachiopods, rhythmites, and stromatolites. Contributions that vary Earthʼs moment of
inertia, such as continental plate drifts, coastline changes, ice age formations, and viscous glacial rebounds, are
superimposed with the secular deceleration. The average deceleration of Earthʼs rotation rate from all available
fossil data is found to be (5.969±1.762)×10−7 rad yr−2. Our value is 99.8% of the total rotational deceleration
determined by Christodoulidis et al., who used artificial satellite data, and our value is 96.6% of the expected tidal
friction value obtained by Stephenson and Morrison. Taking the derivative of conserved angular momentum, the
predicted lunar orbital deceleration caused by the average rotational deceleration corresponds closely to lunar
models. When evaluating the significant time gaps between UT and TAI, Earthʼs rotational deceleration is a minor
contributing factor. Also, the secular deceleration rate is necessary to correctly date ancient astronomical events.
We strongly encourage that more ocean paleontological evidence be found to supplement the record to separate the
many periodic variations embedded in these data.
Key words: ephemerides – Moon – time

1. BASIC CONCEPTS CONCERNING TIME AND EARTH millions of years. A constant deceleration model is an excellent
ROTATIONAL RATE fit to the fossil data. This analysis examines all available
In Deines & Williams (2007), the authors established that paleontological data from corals, bivalves, brachiopods,
rhythmites, and stromatolites in terms of sidereal days that
one contributing factor of divergence between Universal Time
record 360° rotations.
(UT)4 and International Atomic Time (TAI) timescales was due
In this paper, tidal interaction is a cause for the increase in
to the omission of time dilation in the Improved Lunar
the length of the solar day over years, which is one of two main
Ephemeris theory (ILE 1954) that defined time intervals
contributions for the UT time tags lagging behind the TAI
between observed lunar positions. That calibration was
output. Ignoring tidal interactions completely will impact long-
conducted by the US Naval Observatory and the National term operations or calculations. For example, one standard
Physics Laboratory between 1955 and 1958. The scientific value for Earthʼs rotational rate is obtained by equating 2π
community accepted that result (Markowitz et al. 1958) as the radians as the number of UT seconds in a sidereal day
definition for the Système Internationale second. The Deines & (86164.09055+ s) into the total number of seconds in the year
Williams (2007) paper recognized that tidal friction would be a 1900.0 according to Newcomb (1895a, 1895b) and approved
factor that would decelerate Earthʼs rotation rate, but it was not by the IAU (1954). This result is the same value for Earthʼs
analyzed in detail. mean rotation rate in terms of true of date as given by Aoki
The historical record of Earthʼs rotational deceleration is et al. (1982) and in the World Geodetic System 1984 (NIMA
found in all types of ocean paleontology, which this study will 2000) model for the Global Positioning System (GPS).
review. The results of Earthʼs rotational deceleration still show Obviously, Earthʼs rotation rate as of 1958 January was
that tidal effects are initially much smaller than the missing slightly smaller than the derived value of 1900.0, and no
time dilation terms that were published in our 2007 paper, but documentation reveals that any compensation for rotational
botheffects must be included. deceleration was made when TAI was set to equal UT on 1958
The tools and means are there to construct a precise January 0 or in any update to UT since then. GPS uses the same
compensation for rotational deceleration. We will examine 1900 Earth rotation rate today rather than the current Earth
fossil data to get an accurate value for it. Even though rotation rate, as it does not matter since the broadcast GPS
individual fossils are not inherently precise, such accuracy is epoch is always the closest Saturday midnight relative to the
obtained when the samples are spread over hundreds of present with an insignificant difference in the GPS constellation
4
relative to Earth between that GPS epoch and the current time.
UT0 is the timescale based on mean solar time from Earthʼs rotation as It would take decades after an epoch to detect a noticeable
determined by observed meridian transits. Polar wobble affects UT0 at the
observerʼs latitude. The UT1 timescale is adjusted for polar wobble and difference in Earth orientation between the 1900.0 and the
synonymous to UT in this paper. present Earth rotation rate, but the effect could be observed

1
The Astronomical Journal, 151:103 (12pp), 2016 April Deines & Williams

given high enough precision. Another immediate effect of angular deceleration α. Its accuracy is limited to just two
ignoring tidal friction is backdating the calculations for ancient significant figures after subtraction of ω–ω0, where ω = ω0 + α
astronomical events, such as lunar or solar eclipses, because (T–T0). Although decadal periodic variations in Earthʼs
Earthʼs actual rotational hour angle will be oriented further rotation rate interfere with short-term UT measurements, fossils
west than using the 1900.0 rotation rate of Earth. In even earlier provide three significant figures directly from any raw counts of
times, the uncompensated rotational deceleration could accu- growth ridges, despite any fossil record deviations. Averaging
mulate into the loss of a day of rotation in the calculations, as over n samples reduces the inaccuracy, δ, by d n . With
well as affect the Julian Day Calendar accuracy. n > 100, our final result will be four significant figures.
Hence, we choose to use fossil samples containing growth
ridges to determine the average deceleration in Earthʼs rotation
2. TIMESCALE ISSUES, TIDAL FRICTION, AND THE
using an elementary kinematic model. Crustal plate movement
FOSSIL HISTORY
can drastically change the coastlines over many eons. Changes
UT is the timescale based on the mean solar day, which is in Earthʼs moment of inertia can be due to postglacial
divided exactly into 86,400 units called seconds. UT is the rebounding or formation (i.e., glacial isostatic adjustment)
oldest operational timescale that is still used for civil and would also be incorporated in the actual rotation rate
timekeeping, although it does not match the precision obtained recorded many millions of years ago (Mya). Vortices in the
with atomic clocks. The initial value for TAI was chosen so molten lava may modify the moment of inertia on a periodic
that TAI would read the same time as UT at the beginning of basis (Cardin & Olson 1994). A secular change for Earthʼs
1958 January. The two timescales soon began to diverge moment of inertia is not expected, or else Earth would evolve
slowly after 1958, and the difference is called ΔT = TAI − into a flat rotating mass while maintaining elasticity, similar to
UT. TAI is independent of Earthʼs rotation, so that ΔT shows a spinning pizza crust. The fossil record measures all sources of
how UT lags behind the TAI rate. In addition, Mean Sidereal secular and periodic variations of Earthʼs rotation. Throughout
Time (MST; equivalently GMST) is the angle between the this paper, the terms tidal friction and secular rotational
meridians of Greenwich and the vernal equinox. The tropical deceleration are interchangeable.
year (equinox to equinox) can be divided into mean solar days
with reference to the Sun or mean sidereal days with reference
2.1. Evaluation of Paleontological Reports on Earth’s
to the distant stars; the tropical year has a length of exactly one
Rotation Rate
more rotation in sidereal days than in solar days. As Earth
revolves around the Sun in each orbit, one sidereal rotation is Growth rings on fossils and sediment layers caused by the
divided up evenly and incorporated in the mean solar days, so rhythmic action of the tides have the potential to obtain a
the annual sidereal days exceed the solar day count by exactly 1 realistic measure for the magnitude of the deceleration on
day. The tropical year is chosen among other annual periods, Earthʼs rotation independent of timescales. Comparisons of
since it has more precision in its period than others (see paleontological annual records are made between several
Explanatory Supplement by Seidelmann 1992). stromatolites (McGugan 1967; Pannella 1972a, 1972b; Vanyo
In our earlier paper (Deines & Williams 2007), we & Awramik 1985; Sonnet & Chan 1998; Zhu & Huang 2003),
documented that UT, which is based on Earthʼs rotational corals (Wells 1963, 1970, Scrutton 1964, and Pannella 1972a),
rate, always trailed TAI, because the size of the mean UT bivalves (Berry & Barker 1968; Pannella 1972a, 1972b, 1976),
second is minutely larger than the atomic second. This is one of brachiopods (Mazzullo 1971), and rhythmites (Williams 1989a,
many issues that the timekeeping community has not addressed 1989b, 2000; Sonnet et al. 1996). The length of the solar day
adequately, and evidence from Essen et al. (1958) and (LOD) has varied in these geological records from 375 to 532
Markowitz (1959) documented that the average UT second of days per year. The solar day that defined UT is gradually
1958 was larger than the Ephemeris Time (ET) second, which becoming longer over the eons, and the number of solar days
will be discussed in detail in a future paper. per year is getting fewer. Thismeans that the UT second (being
Some critics may contend that atomic clocks are far more exactly 1/86,400 unit of a mean solar day) is also getting
precise than the fossil record. As rotational deceleration is very longer, which is why an epoch such as 1900.0 was picked for
small, it will take centuries to get a valid estimate. As explained the epoch year to define the “standard” UT second.
in this paragraph, atomic clocks can only achieve two Before beginning any model of Earthʼs past rotation rates, it
significant figures in the last century for Earthʼs rotational is necessary to look for a time unit that is not changing. As
deceleration. Earthʼs rotational rate ω of 365.2421897 days in Earth decelerates, the length of the day noticeably increases.
tropical year 1990 (Seidelmann 1992, p. 698) is probably Thismeans that neither the LOD nor the mean solar second for
determined from the atomic timescale after eliminating the UT, which is 1/86,400 of a solar day, can be used as a uniform
periodic variations in the length-of-day (LOD) observations. time unit without specifying an epoch.
Unfortunately, the rotational periodic variations near the 1958 Earthʼs orbital period seems to have no secular change, as
epoch of TAI were not measured or monitored, since no long- suggested by numerical integrations over millions of years and
term clock with better precision than Earthʼs rotation existed by literal expansions of algebraic theories that have not
then. The only other estimate of Earthʼs rotational rate with at produced evidence for secular changes in the semimajor axis to
least the same atomic precision is obtained from centuries of at least third order (Simon et al. 1994). By Keplerʼs third law, a
observations to fix Earthʼs orbital frequency in the formula for stable semimajor axis of the orbit ensures a stable orbital period
Earthʼs tropical year at 1900.0 (Seidelmann 1992, pp. 79–80), in the absence of external forces. Also, if Earthʼs entire
which one obtains by converting the number of seconds in rotational momentum is transferred into its solar orbit, the
tropical year 1900.0 into days (divide 31,556,925.9747 by change in the new period would be less than 15 s. Gravitational
86,400). Obtain the difference between these two angular planetary perturbations to Earthʼs orbit cause the tropical year
velocity rotations and divide by 90 yrto get the resultant to vary from one year to the next, but those perturbations

2
The Astronomical Journal, 151:103 (12pp), 2016 April Deines & Williams

Table 1
Paleontological Results of Earth Rotation in Solar days yr−1

Source Samples Eons Rotation Rate α


Millions of Years Ago Solar days yr−1 Results 10−8 rot yr−2

Pannella (1972b) 2** −1850 442 −4.149


Pannella (1972a) 3** −1750 532 −9.529
Zhu & Huang (2003) Many −1300 546 −13.904
Sonnet & Chan (1998) 1385** −900 464±13.5 −10.973
Pannella (1972a) 1** −900 416 −5.640
Vanyo & Awramik (1985) 4** −850±50a 445.75±32 s.d.b −9.472
Williams (2000) Many −620c 400±7 −5.606
McGugan (1967) 1** −525 410±10d −8.525
Pannella (1972a) 1** −510 442 −15.051
Mazzullo (1971) 1 −425 421 −13.120
Wells (1970, pp. 3–9) 1 −425 412 11.002
Mazzullo (1971) 2 −415 419e −12.954
Wells (1970, pp. 3–9) 1 −415 400 −8.375
Mazzullo (1971) 6 −375 409±3 s.d.f −11.669
Scrutton (1964) 1 −375 401 −9.535
Wells (1970, pp. 3–9) 12 −373 398±6 s.d.g −8.782
Wells (1970, pp. 3–9) 1 −328 398 −9.987
Pannella (1972a) 1 −360 405.5h −11.183
Pannella (1972a) 3 −340 398.3±0.5 s.d.i −9.723
Wells (1970, pp. 3–9) 2 −294 386±4j −7.061
Pannella (1972a) 3 −290 383.5±4.8 s.d.k −6.296
Pannella (1972a) 3 −220 371.6±6.1 s.d.l −2.890
Berry & Barker (1968) 8* −70 370.3±2.2m −7.225
Pannella (1972a) 3* −70 375.1±3.9 s.d.n −14.083

Result using All Data Average E-8 rot yr−2 −9.447


(All symbols in Figure 1) Std. Dev. E-8 rot yr−2 3.196
Acceleration E-8 rad yr−2 −59.358

Result using reports with Average E-8 rot yr−2 −9.693


six or more samples Std. Dev. E-8 rot yr−2 3.063
(Square symbols Fig. 1) Acceleration E-8 rad yr−2 −60.095

Result from Fossil Data Unweighted Average E-8 rot yr−2 −9.592
(shaded table) Std. Dev. E-8 rot yr−2 2.962
(plotted in Figure 2) Acceleration E-8 rad yr−2 −60.270

Result from Fossil Data Weighted Average E-8 rot yr−2 −9.377
(shaded table) Std. Dev. E-8 rot yr−2 3.336
(plotted in Figure 2) Acceleration E-8 rad yr−2 −58.918

Notes.
*

**
Result obtained indirectly from conversion of lunar to annual counts.
Special stromatolite processing.
a
Dated uncertainty is given in report without justification.
b
Reported samples were 409, 435, 454, and 485 days yr−1 with mean and s.d. given.
c
Earlier papers listed 650 Mya, but boundary changed from 570 to 540 Mya, resulting in late Precambrian date change to 620 Mya for rhythmites. Longest
continuous laminae reported by Williams is 60 yr.
d
Original abstract listed 400–420 days yr−1, so midpoint is reported here.
e
Reported two samples with each being 419 days yr−1, so no uncertainty is given.
f
Six reported samples were 403, 409 × 2, and 410 × 3.
g
Twelve graphed dots showed 384, 390 × 2, 396, 399, 400 × 4, 402, 403, and 406.
h
This sample was over 6 yr old showing six sequences, but individual years were unreported. Individual samples extended over n years will be shown as (n).
Reported three samples of 399(4), 398 (9), and 398 days yr–1(5).
i
j
Graphical data displayed two samples of 382 and 390 days yr−1.
k
Reported three samples of 389(3), 381.5 (5), and 380 days yr–1(5).
l
Reported three samples of 375.3 (3), 364.5 (2), and 375 days yr–1(2).
m
Report gave lunar mean of 29.65 and uncertainty of 0.18, which is converted to an annual value by multiplying the annual projection by 0.18/29.65.
n
Reported three samples of 379 (4), 375 (1), and 371.2 days yr–1(4).

3
The Astronomical Journal, 151:103 (12pp), 2016 April Deines & Williams

average out over the long term, especially over millennia. form more layers. The environmental events, both random
Therefore, we adopt the mean year as a consistent unit of time. storms or erosion and cyclic happenings, tend to make
The following equations are used to construct Table 1 from stromatolites less precise as calendars than bivalves or other
the paleontological records: invertebrates. One should expect gaps in the stromatolite
record, but full cycles may be produced if enough samples are
w = w 0 + a ( T - T0) (1 )
interleaved together to get the whole record. Stromatolites tend
1 to grow in columns owing to oozing of the inert material being
q = q0 + w 0 ( T - T0) + a ( T - T0)2 (2 ) layered with algae. Another sediment called rhythmites has
2
been found, consisting of siltstone and fine sandstone deposited
where in layers to form rock strata, probably caused by tidal ebbing
θ = Earthʼs rotational angle in rotations at time T in years, and flowing. Studies involving rhythmites and stromatolites
θ0 = Earthʼs rotational angle at the epoch time T0 in years, usually involve meticulous processes. Some stromatolites were
ω = Earthʼs angular velocity in rot yr−1 at time T, photographed as the cores were polished repeatedly while
ω0 = Earthʼs angular velocity in rot yr−1 at the epoch time shaving off a thickness of tenths of a millimeter, effectively
T0, and creating pictorial slices. This process destroys the specimen,
α = Earthʼs constant angular acceleration in rot yr−2. but it preserves a full three-dimensional record that can be
reexamined as necessary.
Table 1is different from the graph of the tidal friction data Much of the pertinent literature cites raw data from earlier
presented by Deines & Williams (2009) that was based heavily papers, making it hard to find any new listed data and avoid
on Wells (1963), which caused geophysicists and paleontolo- duplication. Some articles are not included in the references as
gists to examine fossil growth rings for determining Earthʼs no new information was given, such as Eicherʼs review (1976)
annual rotation rate eons ago. Wells (1963) graphed the of Wells’s (1970, pp. 3–9) paper or Lambeckʼs reviews (1978,
astronomical dictum that Earthʼs rotation decelerates at the rate 1980). Such articles are extremely useful to educate the novice
of 2 s every 100,000 yr(i.e., the LOD is increasing by in the difficulties that experts encounter inreducingfindings to
20 μs yr−1). Wells (1963) only described two data points from a simple number of revolutions after millions of years (Mya).
fossils, which were not included in his Figure 1. Wells (1970, Lambeck (1980) lists 16 extra fossil records from Mazzullo of
pp. 3–9) corrected the deficiencies with more data in the final the Middle Devonian with an annual average of 405 days and
summary. an extra fossil from the Silurian era than those listed in Table 1.
The intent of this section is to examine many available
Mazzullo provided Lambeck this additional fossil information
paleontological records to include in Table 1 for the grand
through a personal communication in 1977, as noted in
average as the best estimate of Earthʼs rotational deceleration.
Lambeck (1980). We prefer to use the original published data
Early paleontological papers dealing with geochronometry
in the literature for Table 1. Also, most sources cite only the
concentrated on corals, bivalves, and brachiopods, which have
strata in which the specimen was found without spending
general daily growth increments that indicate annual variations.
With successive alteration of daylight and darkness, growth resources, such as radioactive testing, to date the fossil. In those
occurs faster during daylight. Annual variations could be due to cases, the geologic timescale figure from Kulp (1961) was used
changes in daylight length, water temperature, or even food to estimate a date for the specimen or specimens in Table 1.
supply. Shorter periods are found within the annual variations, Several times, authors listed a different result for a
which can imply seasonal fluctuations or lunar synodic months. referenced paper than was originally published in the reference.
Variability in specimen growths was verified in recent animal We assumed that this was a typographical error, and the
growth tests. Although the fossil record was reliable, it did original result was used. One such case deserves mentioning.
require many samples or long, continuous growth histories to McGuganʼs 1967 paper from the New Orleans meeting is only
get better estimates for Earthʼs rotation rate in ancient times. found as an abstract, which states that the stromatolite findings
Critics pointed out the inaccuracies of the raw fossil record, were between 400 and 420 increments from the Upper
which were found later. Many factors, including environment Cambrian age. This was put into Table 1 as the average of
changes, weather, erosion, disease, personal bias, etc., could 410±10 solar days at 525 Mya. Pannella, a meticulous writer,
alter the physical growth patterns and the counts. There is wrote, “McGugan (1967) had published a figure of 424 days
general agreement that the growth rate varies annually in corals for the length of the Middle Cambrian year.” Pannella could
and shells, but there are many variations and factors. Lambeck have been in the audience of that presentation, or, as implied in
(1980, pp. 363–381) and Scrutton (1978, pp. 154–172) give the acknowledgments section (Pannella 1972a), McGugan had
extensive details and references concerning the various growth personal communications with Pannella well after the con-
patterns and influences. We are unable to provide a critical ference. Although the entry of 424 solar days at 525 Mya for
error analysis owing to the lack of data provided in the original (McGugan 1967) is probably correct, the original listing from
reports. We will provide error estimates that are available in the his abstract is still used to be consistent with the other Table 1
publications, or else we will use standard deviations (s.d.) to entries.
estimate the distribution of errors. Many of the results in Table 1 have theoretical calculations
For earlier eons, the organic sedimentary structures or used to obtain the number of revolutions per year. These
stromatolites (lamina created by alternate layers of algae and particular results have an asterisk placed by the specimens.
sediments compressed into rock columns) can be a paleonto- Often, fossils were used to count the days of the lunar synodic
logical clock. Stromatolites were studied to fill in the period month, and from various formulae, such as the changing rate in
before fossils were deposited. The assumption is that cyclical orbital dynamics, the number of months per year can be
rhythms are preserved as algae grow in thin mats or layers, trap calculated. Growth increments in bivalves are affected when
and bind daily sediment deposits, and repeat the process to low tides uncover the shell, which inhibits growth. The daily

4
The Astronomical Journal, 151:103 (12pp), 2016 April Deines & Williams

bands can be divided in two where semidaily tides exist. In specimens. Out of the hundreds of stromatolite specimens,
temperate latitudes, the cooler winter season inhibits growth in specific layers with complete lamina have to be found (often at
corals and bivalves so that the annual cycle can be identified in other sites) to complete the actual record (see Table 2 of
the band patterns. Pannella 1972b). Hence, many papers on stromatolites used
Two highly theoretical papers on corals must be qualified. this technique of constructing two ratios and multiplying them
Scrutton (1964) wrote, “Extrapolating, therefore, from the together to get a truer count of days per annum. These results
astronomical calculations alone, the Middle Devonian year was were given a double asterisk in Table 1.
approximately 399 days in length.” Even Williams (1989b) Interestingly, Kahn & Pompea (1978) documented that the
wrote, “I have not plotted the value of 399 days per Middle nautilus shells have ridges in each chamber, which might be
Devonian year used by Scrutton (1964, 1978) which is not useful for geochronometry. Recent nautilus shells have 29–30
based on coral data but is, in fact, an estimate of the Earthʼs ridges in each chamber (about the number of days per lunar
rotation rate during the Middle Devonian based on astronom- month), and the ridges decrease per chamber with older fossil
ical calculations of the present deceleration of the Earthʼs shells. The results formed a monotonically increasing cubic
rotation.” However, Scrutton did write that specimen OUM curve from the earliest shell fragments of 420 Mya to the
DT2 had 401 ridges between the two edges at two swells. So, present. The earliest nautilus fragments show nine ridges per
Scruttonʼs written result is replaced in Table 1 by the single chamber. A simple conclusion might be made that the Moonʼs
coral finding of 401 solar days at 375 Mya. Based on Scruttonʼs orbital period was 9 days about 420 Mya. With the solar day
calculations, Johnson & Nudds (1975, pp. 27–42) extrapolated being shorter then, this would make the past lunar orbit
the supposed 13.01 synodic months in the Devonian year to approach 8 present days in length. This is not supported by
12.95 months in the Lower Carboniferous (Viséan) year that main-line research, such as documented by Pannella et al.
they were analyzing, and they calculated that the Viséan year (1968), that the lunar period was 31 ancient days long about
was 391.09 days long. This result is so precise that it raises 500 Mya. One alternate explanation is that the nautilus slowly
immediate doubts. Johnson and Nudds used 10 coral samples evolved by changing its growth pattern within its chambers
to get a mean of 30.2 days for the lunar synodic month in the from nineridges per lunar month when it first lived to 30 ridges
Lower Carboniferous eon. However, they cite both Wells each month. Regrettably, the nautilus data are not used for our
(1963) and Scrutton to get their extrapolation for the linear study.
relationship of lunar months per year in ancient times. They
claimed they hadclear evidence that “the number of days per 2.2. Determination of Rotational Deceleration from
Paleontological Evidence
year is decreasing and that the Earthʼs rotation about its polar
axis must be slowing down.” Unfortunately, they didn’t realize Table 1 includes all valid reports found on the subject of
that the astronomical dictum used by Wells guaranteed the Earthʼs historical rotation history. Where a time span was given
obvious outcome. Therefore, their paper was not included in in the paper, the midpoint was used for the entry, such as
Table 1. McGuganʼs time span of 400–420 Mya being replaced with
Pannella (1972a) also wrote, “Mazzullo (1971) did publish a 410 Mya in Table 1. When multiple specimens were listed, the
list of his counts but did not give the important information of average of days per year from all specimens of that eon as listed
how many yearly sequences he counted for each of the 50 in the report was inserted instead of the preferred specimen,
specimens.” However, Pannella (1972a) did include the such as was done with the four specimens in the Vanya and
number of sequences (i.e., yearly growthpatterns per speci- Awramik paper. Wells’s (1970, pp. 3–9) results, which
men) in his reports, which are annotated in Table 1. included his earlier results, are taken from that paperʼs graph.
The mathematics behind stromatolites is simpler and set up In the effort to extract the graphical data, possible interpolation
in ratios from pattern recognition from direct counts of lamina. errors may be embedded in the tabular entries. The updated
Pannella (1972b) defined the five classes of periodic patterns in result of Sonnet & Chan (1998) for the 900 Mya era was
stromatolites as follows: first-order daily lamination (I) inserted in Table 1, rather than the Sonnet et al. (1996) report.
contains the daily deposit (basic bipartite 24hr unit, made of For the same reason, the Pannella (1972b) report was chosen
clastic [corpuscular]and an organic-rich layer), second-order for its results of 900 and 510 Mya eras over Pannella (1972a),
tidal band (II;period of 14.8 days made of two groups of which avoids duplication. The data are converted using
lamina with different thicknesses), third-order monthly band Equation (1).
(III; period of 29.5 day succession of two second-order bands, We require that the intercept of a straight line must go
one better developed and generally less incomplete than the through 366.2421987 days = ω1900 for the epoch T1900. The
other), the fourth-order seasonal band (IV; latitude-dependent present rotation rate is more precise than any of the fossil data.
band consists of several second-order bands grouped in zones, This leaves only the slope of the line as the single degree of
often detected visually owing to color differences), and the freedom in a least-squares fit of the linear equation yi = m
fifth-order annual band (V; period of 365 days that the unit is (T – T1900) + 366.2422, where yi is the number of sidereal days
made of two fourth-order bands, one organic-rich and dark, the per year from the fossils and Ti is in Mya (annual seasonal year
other organic-poor and light). Ratios are made by counting the is closest to the tropical year). Define the ith residual, ri, and the
number of I-order lamina per II-order bands, and by counting sum of squares, S, to be
the number of II-order per V-order bands. Pannella showed that n
direct counting of I-order lamina per V-order band in long ri = yi - m ( Ti - T1900) - 366.2422 and S = åri2.
continuous specimens resulted in gross undercounting of days i=1
per year, because the growth layers are often interrupted. For
example, subaerial exposure interrupts growth that shortens a To find the minimum sum of squares for the slope m, take the
complete cycle, which can happen in long, continuous derivative of S with respect to m and set the derivative to 0.

5
The Astronomical Journal, 151:103 (12pp), 2016 April Deines & Williams

Figure 1. Effect of tidal friction on Earth rotation. Reports with <6 specimens
are plotted with diamonds;reports with …6 specimens are plotted with squares. Figure 2. Earthʼs rotational deceleration from fossil growth rings.

Rearranging the resulting terms, the equation is have been many ice ages and global thaws, so glacial
rebounding is expected to have much shorter periods. More
å ( Ti - T1900) Yi fossils will be needed to fill in the gaps and provide more
m= .
å ( Ti - T1900)
2 historical details in Earthʼs rotation rate.
If one speculates that the solar count is off by +5 days or if
It is seen that the least-squares fit with the requirement that the an eon is off by +25 million years, the respective error can
intercept be 366.2422 sidereal days will result in weighting the drive the result off by 15%or −6%, respectively, if applied to
oldest fossil more than the most recent fossil data, which we the Wells data (circa −373 Mya). Roughly, an error of 1 day is
will not accept. Thus, we will use the mean, rather than a least- about 3% or a 4 million year error is about 1%. Thus, the
individual paleontological data can give reasonable accuracy
squares fit, for equally weighted data to estimate the slope. The
for tidal friction, and averaging would further reduce offsets to
data have three significant figures, and the uncertainty, δ, is obtain the best statistical estimate.
reduced statistically by d n . With n > 100 samples, the final If one selected only those reports with six or more specimens
average will be given in four significant figures. The (six reports shown as squares instead of diamond markers in
uncertainty will be expressed as the s.d., because most reports Figure 1), the average rate of deceleration is −9.693E-8
fail to give uncertainties with the raw data. The significant compared to the overall average of −9.447E-8 rot yr−2. The
figures deal with precision, not accuracy. fossil data shown in Figure 2 are sufficient to determine the
Table 1 includes every actual observation to obtain the linear effect of tidal friction on Earthʼs angular rotation. Using
overall average. For consistency, the solar data are converted only the fossil data and treating each row as an independent
into sidereal days, representing the number of 2π radian entry, the average deceleration is −9.592E-8 rot yr−2, with an
rotations counted per year. s.d. of 2.962E-8 rot yr−2, giving an acceleration of
A graph of the tabulated data is shown as Figure 1. The slope (−60.52±18.61)E-8 rad yr−2. We included the Berry &
is the average of the individual deceleration rates obtained from Barker (1968) report for that unweighted average, but excluded
the plotted data shown in both Figures 1 and 2. it when calculating the weighted average as it had eight
Table 1 shows the final results of various paleontological specimens. Berry and Barker carefully measured the ridges in
reports based on single or multiple specimens. Most reports did the bivalve specimens between high tides, which is between
not provide enough information to weight the data, such as the full Moon and new Moon phases per fortnight. They obtained
quality of the specimens, the total continuous length in years of 24.98±0.04 ridges, which they interpreted that 12.49 lunar
each specimenʼs record, or how some of the values were months per year occurred in the Late Cretaceous year. This
obtained. The model used for tidal friction is a simple constant means that the Moon had a larger orbit 70 Mya, which shrank
rotational deceleration, which will obtain a mean constant slope to the present 13.37 orbits per year and is very slowly
for the straight line. So, we computed an average for the increasing owing to the transfer of Earthʼs rotational angular
constant rotational deceleration using the data and Equation (2). momentum into the lunar orbit. No long-term lunar theory
The slope is the average of the individual decelerations supports this scenario. However, if one assumes that the
obtained from the data in each figure. bivalves grow ridges when the Moon is visible at night, there is
Figure 1 confirms that the best estimator for Earthʼs rotation about one day on either side of the new Moon at nearly the
rate is the normal mean when many independent samples are same meridian as the Sun that would make the Moon invisible
available to measure its effects. Elimination of the oldest outlier at dusk between waning and waxing lunar crescents. Add that
result could be tested by statistical criteria. The sedimentary gap to 24.98 to get a realistic lunar period. Berry and Barkerʼs
deposits have larger variations from the straight line, and fossil annual day estimate is too pessimistic, so we will not include it
data seem to indicate a more linear trend. That could be the in the weighted fossil average owing to that reportʼs large
resultof incomplete stromatolite records, because it is more number of specimens. The weighted average rotational
difficult to find such ancient samples at different sites scattered deceleration is (−9.377±3.336 s.d.)E-8 rot yr−2 when adjust-
over various latitudes. The fossil data do show some cyclical ing all the fossil data by the number of specimens.
variation, which could be due to the moving tectonic plates that Using the number of seconds in the tropical year at 1900.0,
change the coastlines over hundreds of millions of years. There the weighted average rotational deceleration from fossil data is

6
The Astronomical Journal, 151:103 (12pp), 2016 April Deines & Williams

astronomical observations performed over


the interval from 1750 to 1892
(Newcomb 1895a, 1895b). Consequently, the
second of ET had the same duration as a second
of UT1 that would have been observed in about
1820, the approximate mean epoch of the
observations analyzed by Newcomb. (Note,
incidentally, that 1900 was the epoch of a tro-
pical year of 31,556,925.9747 s of ET, while
1820 was the epoch of a LOD of 86,400 s
of UT1.)

Their comments about 1820 are based on a parabola


Figure 3. Geometry of Earthʼs spin axis relative to lunar orbital plane and
ecliptic. published by Stephenson & Morrison (1995) and by McCarthy
& Babcock (1986, p. 281), which spanned data between 1656
and 1984. An epoch in any timescale is a chosen time instant
equivalent to (−5.916±2.105 s.d.)E-22 rad s−2. This is within for reference, not an average date from a time span of years.
1.1% of the total tidal deceleration (−5.98±0.22)E-22 rad s−2 The year 1900 is the epoch for UT, formerly called Greenwich
found by Christodoulidis et al. (1988), who used precision laser Mean Time (GMT), in Newcombʼs formula, which is based on
and Doppler range data from artificial satellites. The team used Earthʼs rotation rate. Otherwise, any mean solar day is exactly
the GEM-TI model to simultaneously recover a complete 86,400 UT1 seconds by definition, so that any year can be an
gravitational field to degree and order 36 and to solve 66 tidal epoch of UT1 in UT1 seconds. The epoch fixes the size of the
coefficients representing the longest wavelength coefficients of UT1 second relative to the year, because Earthʼs rotational
12 major tides with 550 other ocean tidal terms of 32 major and deceleration causes fewer days per year over eons. UT was the
minor tides. In all, 17 satellites were used to collect precise only timescale used until Clemence proposed the concept of ET
range data taken from seven third-generation lasertracking in 1948, which was later approved formally by the IAU in
systems (<5 cm) and Tranet Doppler from Seasat and Oscar 14 1960. 1820 is simply the midpoint of the selected data span,
satellites. The data were plotted in an Earth equatorial system in and their analysis is explained by a constant angular
the GEM-T1 model. The constants used to generate the normal deceleration of Earthʼs rotation.
equations for the GEM-T1 geopotential model are listed in that The epoch for Newcombʼs formula can be calculated
reportʼs Table 4. The perturbed orbital elements provided the independent of fossils. The epoch T0 is computed from
solution through various processes to obtain Earthʼs rotational Equation (1). Earthʼs rotation rate of 7.292115855E-5 rad s−1
deceleration. Our standard deviation of 2.106E-22 rad s−2 is
is obtained from Newcombʼs formula cited above by dividing
larger than the deviation of 0.22E-22 rad s−2 from Christolou-
the number of sidereal days of 2π radians each by the number
lidis et al., because of the long-term periodic variations
of seconds in the standard tropical year of 1900 (refer to Deines
embedded in the fossil data (see Figure 2). If more fossil data
& Williams [2007] for obtaining the seconds in 1900). This
are discovered to complete the paleontological history, one
rotation rate is identical to the one quoted by Aoki et al. (1982)
may be able to analyze the contributions of various changes to
Earthʼs moment of inertia, such as glacial rebound, continental and is identical to the rate still used in the World Geodetic
drifts, etc., over eons. System 1984 (NIMA 2000) in computations made in the GPS.
It is also noted that the average rotational deceleration is The number of solar days in year 1900 is 365.2421987 by
96.6% of the expected −6.203E-22 rad s−2 that Stephenson & dividing the number of seconds in 1900 by 86,400 s in a solar
Morrison (1995) computed for tidal friction. For the remaining day. The Explanatory Supplement of the Astronomical
analysis in this paper, we will choose the average deceleration Almanac (Seidelmann 1992, p. 698) lists the tropical year as
rate obtained from the fossil data. 365.2421897 solar days for 1990, which can be converted to
sidereal days in radian units and divided by the number of
seconds in a standard tropical year to get Earthʼs rotation rate at
3. REEVALUATING THE PHYSICAL BASIS OF THE 1990. Use the deceleration α from artificial satellites as
LEAP SECOND AND THE UT EPOCH calculated by Stephenson & Morrison (1995). Insert these
This new evaluation for Earthʼs deceleration can now be values into Equation (1) to get T0 = 1898. The ω–ω0 term in
applied to check the claims that Earthʼs tidal friction is the sole Equation (1) leaves only two significant digits after subtraction.
basis for the timescale divergence between UT, based on All these values used in Equation (1) are independent of our
Earthʼs rotation, and TAI, based on atomic clocks. For fossil determination of tidal deceleration and are all measured
example, McCarthy et al. (2008) wrote, values of Earthʼs rotational rate that determines UT1. In any
case, the epoch is rounded to 1900 for T0 as the UT epoch,
On the basis of Newcombʼs formula for the which demonstrates that 1820 is not the epoch.
geometric mean longitude of the Sun, the sec- So, why is 1820 the supposed epoch of UT1? The answer is
ond of ET was defined ... as “the fraction 1/ found again by Equation (1). Choose two times TA and TB and
31,556,925.9747 of the tropical year for 1900 the associated rotation rates ωA and ωB. Create two equations
January 0 at 12 hr ephemeris time” (BIPM 2006, by using Equation (1) to replace ω and T at events A and B
p. 149). Newcombʼs formula was derived from with ωA, TA, ωB, and TB, respectively, then add the two

7
The Astronomical Journal, 151:103 (12pp), 2016 April Deines & Williams

equations together and divide by 2 to obtain rotational UT data (T0 = 1900). Newcomb did his work
correctly with the stated 1900 epoch in the formula for the
1 2w 0 2aT0 a
( wA + w B) = - + ( TA + TB) mean motion of the Sun, which produces the equivalent time
2 2 2 2 for UT1. Newcombʼs mean solar position formula defined a
wA + w B ⎛ TA + TB ⎞ smooth GMT (now called a UT1 timescale), which was the
= w 0 - aT0 + a ⎜ ⎟. (3 )
2 ⎝ 2 ⎠ only international timescale used at that time. The epochs of ET
and UT are needed to compute the effect of tidal friction using
Note that the left term in Equation (3) is the numerical average Equation (2), where the angular displacement θ is converted
angular rate of a data span bordered by points A and B. As the into the equivalent of sidereal hour angles (15° = 1 hr).
numerical years are in unitary increments, the right term is both
the numerical average of the span of years and the midpoint. 4. EARTHʼS ROTATIONAL DECELERATION AND
Other terms are constants in Equation (3). Stephenson & LUNAR ORBITAL DECELERATION
Morrison (1995) cited their earlier paper of 1984 on the
One needs to determine whether Earthʼs rotational decelera-
observed lunar occultations as a function of UT between 1656
tion is consistent with other related values, such as the lunar
and 1984. The raw data were ΔT = ET–UT. McCarthy & orbit deceleration values that have been published. Earthʼs
Babcock (1986) published a similar paper using Stephenson rotational deceleration that is obtained from the fossil data
and Morrisonʼs same data, with one exception of Martinʼs should transfer rotational momentum to the lunar orbit by
(1969) data in the 1656–1860 interval, which would offset raw momentum transfer. Our fossil analysis has established a value
data in the following years. The total data spanned 1656–1984 of the tidal deceleration over hundreds of millions of years—far
with a midpoint of 1820 for both papers. Serendipitously, longer than has been accomplished previously. Within the level
Newcombʼs midpoint is 1821 using data from 1750 through of significance obtained for tidal friction, we make the
1892. If the ET epoch is claimed to be 1900, then the raw data following assumptions to simplify the derivation. The obliquity
of ΔT ≈ 0 will indicate the UT epoch. Stephenson & Morrison of the ecliptic is assumed to be the mean value of 23° 26′
21″ = 23°. 43916 for the inclination of Earthʼs spin axis relative
(1995) showed that ΔT was virtually zero at 1900 and zero at
to the ecliptic (Seidelmann 1992). For the next 1 million years,
1900 from their fitted parabola, while McCarthy and Babcock
the obliquity may be between 22° 13′ 44″ and 24° 20′ 50″
showed that the raw data are zero in 1902. In McCarthy and (Berger 1976), but such variations will not degrade the
Babcockʼs paper, ΔT = +11.33 s at 1820 in the raw data, and calculations in the cosine projection of Earthʼs rotational
the fitted parabola showed −20.0 s at 1820 for ΔT = ET– angular momentum to the ecliptic over hundreds of millions of
UT;neither are close to zero. Again, 1820 is not a likely epoch years. We will not consider tidal frequency components or
for UT. Doodson numbers to model tidal friction, which requires
Both Stephenson & Morrison (1984) and McCarthy & extensive knowledge of the amplitudes of each component in
Babcock (1986) will get a symmetric fitted parabola, because the Fourier series analysis. Such tidal effects are already
both used the average angular velocity (rate that Earth spins) included in the fossil data, making tidal models unnecessary in
between the years 1656 and 1984. If the angular deceleration, this analysis. Some may argue that the eccentricity of the lunar
α(e.g., tidal friction), is essentially a constant, then the angular orbit induces tidal dissipation that affects the longitude
velocity changes linearly (i.e., ω = ω1820 + α{T – T1820}) from evolution, but our estimates justify that we can ignore this
maximum at 1656 to minimum at 1984. It is an easy proof to contribution. For example, the mean lunar eccentricity is
show that the average angular velocity, ω1820, is the same at the 0.05490 with a 3232 day period of revolution of the lunar
midpoint and median when α is a constant. Integrating the ω perigee (Seidelmann 1992), so that this effect averages out over
equation will obtain the angular displacement equation for θ. centuries. We are using the mean motion of the Moon, the
ET incorporates a constant angular velocity where a time is mean distance of the Moon from the barycenter, the mean
assigned by a meridian transit of a star (i.e., θET = θ1820 + distance ofEarthʼs center from the barycenter, the mean
ω1820{T – T1820}), while UT has an additional constant angular inclination of the lunar orbit to the ecliptic as 5°. 145396, the
deceleration (i.e., θUT = θ1820 + ω1820{T – T1820} + α mean inclination of the lunar orbit to the lunar equator as
{T – T1820}2/2). The timescale difference ΔT = ET–UT will 6°. 683, and the period of revolution of the line of nodes as 6798
be ΔT = −α{T – T1820}2/2. So, both papers will have an days = 18.6123 yr(Seidelmann 1992). The derivation will
artificial symmetric parabola for ΔT centered at 1820 when calculate the transfer of Earthʼs rotational angular momentum
using the average angular velocity and using the same time to the lunar orbit. Only long-term effects that occur over a
span of data. century will be needed, as short-term effects like lunar
We conclude that the 1820 claim for the UT epoch is not librations will average out. The derivationʼs purpose is to
valid. Rather, 1820 is simply the midpoint of the same ΔT data obtain a first-order calculation for consistency between our tidal
span in both papers. The fact that the ΔT data and fitted friction result independently obtained from fossils and the
parabolas are symmetric and a minimum at the midpoint is published findings of the lunar orbital deceleration.
characteristic that UT is affected by a constant angular
deceleration and ET assumes a constant angular velocity. If
4.1. Derivation of Lunar Orbit Deceleration
no overlapping subsets of the ΔT data span were selected, the
from Tidal Friction
minimums of the fitted parabolas would probably be at
different midpoints and well displaced from 1820. The raw The lunar tidal friction deceleration will transfer a portion of
ΔT data confirm that the UT epoch is 1900 (ΔT ≈ 0 at 1900). Earthʼs rotational angular momentum into the orbital angular
It is also supported by the independent confirmation from 1990 momentum between Earth and the Moon. Solar tides are

8
The Astronomical Journal, 151:103 (12pp), 2016 April Deines & Williams

roughly half the height of lunar tides, so 2/3 of the total rm=mean distance of Moonʼs center from Earth–Moon
deceleration of Earthʼs rotation should be attributed to the lunar barycenter=3.844×108 m.
tidal deceleration that is transferred to the Earth–Moon orbit.
The balanced moment arm equation at the barycenter obtains
This rough estimate can be refined by comparing the
the value for re:
gravitational tidal forces of the Moon and Sun directly on
Earth, yielding the ratio of a lunar tidal acceleration to a solar MM
tidal acceleration on Earth, which is 2.21 (Olah 2009). For the ME re = MM rm or re = rm = 0.01230002 rm. (5 )
ME
first-order calculations, assume that the Moon is spherical and
uniform in density so that its moment of inertia will be 2/5 The next step is to take the derivative of Equation (5), which is
MMR2 with the mean radius R = RM. Assume that Earth has a zero when no external torque L exists. By definition, the
uniform density and is an oblate spheroid spinning about its derivative of Earthʼs rotation rate is the total deceleration α of
semiminor axis. This moment of inertia is 2/5 MER2, where R which the lunar contribution is 2.21/3.21 of the total, or
is the semimajor axis, which is reasonable for a first-order
dw E
calculation. a lunar º - = - 4.056 ´ 10-7 rad yr-2 .
The geometry of the lunar orbit, the ecliptic, and Earthʼs spin dt
axis must be specified before proceeding with the derivation. Take the derivative of Equation (4), set it to zero, and move the
Earthʼs mean obliquity of the ecliptic is 23°26′ first term to the left:
21 448 = 23°. 4392911 (Seidelmann 1992), so that the transfer
of spin angular momentum to Earthʼs orbit would be a cosine 2 dr
projection of 23°. 439. The lunar mean inclination to the ecliptic - a ME RE2 cos ( IE ) = 2we ME re e
5 dt
is 5°. 145396 (Seidelmann 1992), but the lunar line of nodes has drm d w
a precession of about 18.6 yr. The Moon rotates about its spin + 2w m MM rm + ME re2 e 
axis on average as fast as its orbital frequency around the dt dt
2 dw M dw
Earth–Moon barycenter, which is also the same for Earthʼs + MM RM cos (IM )
2
+ MM rm2 m
orbital frequency when circling this barycenter. Also, the 5 dt dt
Moonʼs equator is 6°. 683 relative to its orbital plane, so that any ⎛ MM ⎞ drm
transfer of angular momentum is modified by the cosine = 2w m MM rm ⎜ 1 + ⎟
⎝ ME ⎠ dt
projection of the lunar spin axis onto the lunar orbital plane.
With an eccentricity of 0.05409, which is nearly circular, the ⎛2 ⎛ ⎞ ⎛ M ⎞ ⎞ dw
rate of precession of the orbital plane is virtually constant. The + ⎜⎜ MM RM2 cos ⎜ IM ⎟ + MM rm2 ⎜ 1 + M ⎟ ⎟⎟ m . (6 )
⎝5 ⎝ ⎠ ⎝ ME ⎠ ⎠ dt
perpendicular vector to the lunar orbital plane varies between
18°. 294 and 28°. 585 relative to Earthʼs spin axis with an overall Using Keplerʼs third law to relate mean motion, n, and orbital
average of 23°. 439. Also, the lunisolar precession causes semimajor axis, a ≈ rm,
Earthʼs spin axis to move in a complete cycle in roughly 26,000
yr, while maintaining the same inclination to the ecliptic. a 3n2 = G ( ME + MM) 4p 2, (7 )
Figure 3 shows the orientation of Earthʼs spin axis relative to −1
the ecliptic and the precession of the lunar orbital plane. Over one gets nM+E = 2.665309E-6 rad s for Earth and the Moon
the long term, the average contribution from Earthʼs transfer of orbiting about their mutual barycenter. Take the derivative of
rotational angular momentum to the lunar orbit will be a cosine Equation (7) and rearrange to get
projection of 23°. 439. n˙ a˙ 3n
The total angular momentum (H) equation is 2 = -3 , or n˙ = - a˙ . (8 )
n a 2a
(
H = IEarth w E cos ( IE ) + ME re2 we ) In Equation (8), n = wm = we , a = re +rm ≈rm to four
dw m drm
( ( )
+ IMoon w M cos IM + MM rm2 w m ) (4 ) significant figures, n˙ =
dt
, and a˙ =
dt
. Substitute Equa-
tion (8) into Equation (6)and solve for the radial derivative:
where
drm
ME=Earth mass=5.9742×1024 kg, a˙ = =
dt
MM=Moon mass=0.0123002 ME, 2
ωE=Earthʼs sidereal rotation rate = 366.2421987 rot yr−1, - ME RE2 cos (IE ) alunar
5
ωM=ωe=ωm=Moonʼs sidereal orbital frequency ⎛ MM ⎞ 3 wm ⎛ 2 ⎛ M ⎞⎞
= 2π/27.32166 days = 84.225029 rad yr−1, 2w m MM rm ⎜ 1 + ⎟- ⎜ MM RM cos (IM ) + MM rm2 ⎜ 1 + M ⎟ ⎟
2
⎝ ME ⎠ 2 rm ⎝ 5 ⎝ ME ⎠ ⎠
I= moment of inertia of uniform rotating spheroid of mass
2 = 3.005 cm yr -1.
M and semimajor axis R= MR2, (9 )
5
IE=Earthʼs average inclination to Earth–Moon orbital Substitute the result from Equation (9) into Equation (8) to
plane = mean obliquity=23°. 439,
calculate n˙ = -9.8610E - 9 rad yr-2 . The tidal friction stan-
IM=Moonʼs spin inclination to Earth–Moon orbital
plane=6°. 68, dard deviation using only fossils is 35.6% of the average, so the
RE=Earthʼs mean equatorial radius=6,378,137 m, s.d. for our calculation of ṅ is the same percentage, which will
RM=Moonʼs mean radius = 1,738,000 m, obtain the result of n˙ = (-20.34  7. 24) cy-2 . This range
re=mean distance of Earthʼs center from Earth–Moon includes the original ILE lunar deceleration of -22. 44 cy-2
barycenter=4.728×106 m, that Clemence (1948) computed from Spencer Jones and de

9
The Astronomical Journal, 151:103 (12pp), 2016 April Deines & Williams

Sitterʼs analyses for tidal friction and later included as the tidal and is not even half of the reported 32.184 s gap between TAI
term in the lunar longitude correction of the ILE (see Section 2 and UT1 at 1958.
ofDeines & Williams [2007] for these details about the ILE). Consider another offset between TAI and UT1. TAI was
This range of values for ṅ also encompasses the previously defined with its epoch of 1958 January 0 with the same starting
mentioned estimates for the lunar orbital deceleration between value as UT1, but TAI began to run faster than UT, such that it
was 10 s ahead by 1972 January. The secular contribution in
−24″ cy−2 and −26″ cy−2. The rate of change in the orbital
Equation (2) of 0.5α(T–T0)2 between 1958 and 1972 would be
radius from Equation (9) is slightly less than the 3.79 cm yr−1
0.79 s. It would take 15.7 yrfor a 1 s gap to appear between
rate often cited (e.g., Williams & Dickey 2002), but it is within TAI and UT1 after 1958. Some added effect must be causing
one standard deviation. In any case, the fossil evidence has this divergence between the TAI and UT timescales that is 10
provided an independent, accurate determination of Earthʼs times larger than Earthʼs deceleration over this short period of
total tidal friction, which is consistent with the published lunar 14 yr.
orbital deceleration calculations. The standard deviation from
the fossil data may be due to periodic variations that are 6. EFFECT OF TIDAL FRICTION WITH THE JULIAN DAY
superimposed over the rotational deceleration, which is CALENDAR
apparent in Figures 1 and 2. The lunar portion from the tidal
friction corresponds closely to the lunar tidal correction Astronomers commonly use the Julian Day calendar that was
proposed by Joseph Justus Scaliger in 1582 to simply count
incorporated in the ILE, which was used to calibrate the
days from an early epoch so that all recorded history would be
atomic oscillators to ET. More sophisticated models (e.g., encompassed by it. The Julian Day calendar has an epoch of
Earthʼs and Moonʼs geoid anomalies beyond the respective 4713 BC, which is the result of synchronizing the three main
oblate or spherical shape, nonuniform densities, elliptical calendars of his time. The underlying assumption is that Earthʼs
orbits, librations, perturbations in orbital parameters, preces- rotation rate remains steady over thousands of years, which was
sion, nutation, geological changes in the coastlines, continental the conventional thought of the sixteenth century. Immanuel
drifts, etc.) should still yield roughly the same results at this Kant (1754) proposed that tides may be responsible for
level of precision. One conclusion is thatthe ILE incorporated gradually slowing Earthʼs rotation rate. No one seriously
the lunar orbital deceleration of −22 44 cy2, which is very considered that concept until Adams (1853) and later Delaunay
close to −20 34 cy−2 by the first-order tidal calculation (1859) documented that Earthʼs rotation rate had a secular
affecting the lunar orbital deceleration. Earthʼs rotational deceleration, which was the indirect result of the observed lunar
orbital deceleration that was left unexplained by Laplaceʼs
deceleration was adequately incorporated in the ILE model of
theory.
the Moon used to calibrate TAI, so hardly any tidal friction
The problem is that no single society existed from 4713
contribution should remain to cause TAI to run faster BCto the present with a continuous calendar that tallied actual
than UT1. days from sunset to sunset. Any remnants of ancient calendars
are often sketchy and incomplete. The Sumerians were
probably the first people that recorded stellar observations of
5. EXAMINING OFFSETS BETWEEN UT1 AND TAI the Sun, Moon, and stars against everyday activities,with the
We have established that the UT1 epoch is 1900 as based on belief that deities revealed their guiding influences through the
calculations with observational evidence. Newcombʼs formula stars.
for the fictitious mean Sun already used 1900 January 0, 0 h The chronology of the ancient Near East provides a relative
Greenwich mean noon as the epoch (Seidelmann 1992). Even framework of dates for rulers, dynasties, and various events,
ET was set at 0 January 1900 for its epoch, as approved by the but fixing that chronology absolutely is difficult. For example,
the Babylonian king Ammisaduqa was the fourth ruler
IAU General Assembly in 1955 September and again approved
after Hammurabi in the relative calendar, but the eighth year
by the IAU General Assembly in 1958 (Seidelmann 1992).
of Ammisaduqaʼs reign was recorded by the calendar in the
This is further corroborated in the Explanatory Supplement
Venus tablets, which begins with the conjunction of the rise of
(1961, p. 73), where Tu = 1900. In the 1977 IAU General Venus with the new Moon that occurred either at 1574 BC,
Assemblyʼs Resolution 5, Note 2, the first mention of the time 1638 BC,or 1694 BC, depending on whether the
bias of 32.184 s is listed and attributed to the random and short, middle,or long relative chronology is used (Reiner &
systematic errors arising from setting the lengths of the TAI Pingree 1998; Hunger & Pingree 1999). These years were
second. Later, McCarthy et al. (2008) wrote, “A practical probably computed without tidal interaction in Earthʼs rotation
realization of TT (Terrestrial Time) is TT = TAI + 32.184 s,” rate. Apply the angular deceleration of −9.377E-8 rot yr−2 in
without explaining the source of the time bias. This secular the formula θ = α(T)2/2, and tidal friction leads to a gain of
retardation in Earthʼs rotation has been attributed to tidal 0.604 rotations at 1574 BC, at which Venus would have risen
interaction (Seidelmann 1992, p. 77) to explain that UT1 trails above the ancient Mesopotamia horizon 14.5 hr later. This
behind TAI. probably means that the eighth year of King Ammisaduqaʼs
Excluding the variations in Earth rotation rate that average reign is dated incorrectly by several decades.
out, the secular contribution in Equation (2) is the rightmost If one dates such old events by using the modern rotation
term of 0.5α(T–T0)2. Select T = 1958 with α = 9.377E- rate (e.g., 365.2421987 solar days yr–1), then tidal friction
8 rot yr−2 where one rotation is 86164.09 s for a sidereal day. directly affects the extrapolation to the Julian Day calendar. As
The resulting gap between TAI and UT1 from Earthʼs its epoch is 4713 BC, there are 6727 yrthrough 2015 (no year
rotational deceleration after 1900 is a loss of 13.6 s in UT1 is labeled 0 BC). Put the angular deceleration of −9.377E-
trailing TAI. Earthʼs rotational deceleration is much smaller 8 rot yr−2 in the formula θ = α(T)2/2, and tidal friction leads to

10
The Astronomical Journal, 151:103 (12pp), 2016 April Deines & Williams

a loss of 2.122 rotations in 6727 yr. Since the year has a The previous explanation for the divergence between UT1
relatively constant length, the literal extrapolation of days or and TAI had been attributed to tidal friction alone. Two cases
rotations is incorrect technically when backdating from modern in 1958 and 1972 demonstrate that Earthʼs rotational decelera-
times to earlier recorded events without tidal friction. To tion is too small to explain the total divergence between UT
miscount by 1 day, it takes 4618 yr(roughly when the first and TAI.
Sumerian calendar could have been written) if tidal friction is Earthʼs rotational deceleration directly affects dated events
ignored. The count of days is linked to the number of actual based on the solar day. Backdating from the present to ancient
rotations per year, which decreases over centuries. One must be events like eclipses in the Julian Day calendar using a constant
cautious in backdated calculations without tidal friction, rotation rate without tidal friction may cause the time
because any early recorded astronomical event before 2600 calculations to be off by as much as 1–2 days. The astronomy
BC would be off by about 1 day. community may need to explore various options to modify the
Julian Day calendar to eliminate the miscounted days starting
after its epoch at 4713 BC. Also, Earthʼs rotational deceleration
7. SUMMARY is needed to date ancient calendars accurately based on
astronomical observations, like the ancient Mesopotamian
With so many issues related to paleontological evidence, this calendar that documents the rising of Venus in conjunction
section becomes a summary of the many different findings with the new Moon on the eighth year of King Ammisaduqaʼs
contained in this manuscript. We chose the year as the stable reign. We recommend that more fossil samples be collected to
time unit for this analysis. As no observed secular change in improve the historical picture of Earthʼs dynamical changes in
Earthʼs orbital semimajor axis has yet been detected, the year is rotation, because there appear to be long-term periodic
a consistently stable time unit, rather than the UT day or UT variations that may well correlate with continental drifts and
second. Fossil data are used to compute an average for the glacial isostatic adjustments.
constant angular acceleration of Earthʼs rotation due to tidal
friction, which is found to be −5.892E-07 rad yr−2 or −5.92E- The authors gratefully acknowledge the financial help of the
22 rad s−2. This is 99.8% of the total tidal deceleration of University of South Florida Mathematics Department, where
−5.98E-22 rad s−2 found by Christodoulidis et al. (1988), who Dr. Williams was a faculty member for many years.
used precision laser and Doppler range data from artificial Special thanks are extended to Victor Slabinski for his
satellites. It is also 96.6% of the expected value for tidal friction insightful early comments and suggestions, especially with
from Stephenson & Morrison (1995). The graph of the regard toour previous paper, which led to this paperʼs
paleontological data reveals that the constant deceleration reevaluation of tidal friction and its effects on timescales.
model is justified. Limiting the paleontological data to six or
more samples in any report resulted in a slightly more negative
average deceleration (2.5% different) than the grand average REFERENCES
over all paleontological data, which included the sedimentary
Adams, J. C. 1853, Phil. Trans. R. Soc. Lond., 143, 397
deposits. The grand unweighted average has been computed as Aoki, S., Kinoshita, H., Guinot, B., et al. 1982, A&A, 105, 359
−9.447±3.196E-8 rot yr−2, or −5.96±2.02E-22 rad s−2, Berger, A. L. 1976, A&A, 51, 127
which is 0.3% of the deceleration determined from artificial Berry, W. B. M., & Barker, R. M. 1968, Natur, 217, 938
satellites. Fossil data are direct measurements of the actual Bureau International des Poids et Mesures [BIPM] 2006, The International
rotation rates of Earth over many eons ago with all effects System of Units (SI) (8th ed.; Sevres: BIPM)
Cardin, P., & Olson, P. 1994, PEPI, 82, 235
combined (both secular and periodic). The many fossil data Christodoulidis, D. C., Smith, D. E., Williamson, R. G., & Klosko, S. K. 1988,
show thatEarth’sdeceleration rate is nearly a constant, which JGR, 93, 6216
corroborates our assumption concerning tidal friction. Clemence, G. M. 1948, On the System of Astronomical Constants AJ, 53, 169
With this realistic value for Earthʼs deceleration rate, it is Deines, S. D., & Williams, C. A. 2007, AJ, 134, 64
Deines, S. D., & Williams, C. A. 2009, in Time Measurement and Time Scales,
determined that the UT1 epoch is 1900.0 as computed by 40th Division on Dynamical Astronomy (DDA) Meeting, Earth Rotational
Newcomb in his formula for the Sunʼs apparent movement Deceleration Based on Fossil Data
over centuries. This was confirmed independently by other data Delaunay, C. E. 1859, Comptes Rendus Acad Sci Paris Séance du, 44, 817
using Equation (1). It has been shown that the claimed UT Eicher, D. L. 1976, Geologic Time (2nd ed; London: Prentice-Hall)
Essen, L., Parry, J. V. L., Markowitz, W., & Hall, R. 1958, Natur, 181, 1054
epoch of 1820 is nothing more than the midpoint of the data Explanatory Supplement to the American Ephemeris and Nautical Almanac
spans used in other reports. This is probably caused by using an 1961 (Washington, DC: U.S. Government Printing Office)
average Earth rotational rate over the observational database for Hunger, H., & Pingree, D. E. 1999, Astral Sciences in Mesopotamia, Vol. 44
both UT and ET, and the fitted ΔT result would then be a (Leiden: Brill), 32
symmetric parabola with a minimum at the midpoint of the Improved Lunar Ephemeris 1952-1959 (1954) (Washington DC: U.S.
Government Printing Office; prepared jointly by Nautical Almanac
data span. Offices of the United States of American and the United Kingdom)
From the conservation of angular momentum of the Moon International Astronomical Union 1954, Transactions of the IAU General
and Earth, Earthʼs rotational deceleration will cause a lunar Assembly, VIIIB (Cambridge: Cambridge Univ. Press)
orbital deceleration. From first-order calculations, Earthʼs Johnson, G. A. L., & Nudds, J. R. 1975, in Growth Rhythms and the History of
the Earthʼs Rotation, ed. G. D. Rosenberg & SK Runcorn (New York:
rotational deceleration will contribute a lunar deceleration of Wiley)
n˙ = (-20.40  7.26)  cy-2 . This encompasses the original Kahn, P. G. K., & Pompea, S. M. 1978, Natur, 275, 606
lunar deceleration of -22. 44 cy-2 that was included as the Kant, I. 1754, Kants Werke, Bd I, Vorkritische Schriften I, 1747-1756, S. 185
tidal term in the lunar longitude correction of the ILE. It also Kulp, J. L. 1961, Sci, 133, 3459
Lambeck, K. 1978, in Tidal Friction and the Earthʼs Rotation, ed. P. Broche &
includes several other lunar models of orbital deceleration. This J. Sündermann (Berlin: Springer), 145
confirms that this rotational deceleration is consistent with Lambeck, K. 1980, The Earthʼs Variable Rotation (Cambridge: Cambridge
present lunar orbital deceleration values. Univ. Press)

11
The Astronomical Journal, 151:103 (12pp), 2016 April Deines & Williams

Markowitz, W. 1959, AJ, 64, 106 Reiner, E., & Pingree, D. E. 1998, Babylonian Planetary Omens, Vol. 3, 1
Markowitz, W., Hall, R., Essen, L., & Parry, J. V. L. 1958, PhRvL, 1, 105 Volume 2 of Bibliotheca Mesopotamica Babylonian Planetary Omens
Martin, C. F. 1969, PhD diss., Yale Univ. Scrutton, C. T. 1964, Palaeontology, 7, 552
Mazzullo, S. J. 1971, Bull. Geol. Soc. Am., 82, 1085 Scrutton, C. T. 1978, in Tidal Friction and the Earthʼs Rotation, ed.
McCarthy, D. D., & Babcock, A. K. 1986, Physics of the Earth and Planetary P. Brosche & J. Sündermann (Berlin: Springer)
Interiors, Vol. 44 (Amsterdam: Elsevier) Seidelmann, P. K. (ed.) 1992, Explanatory Supplement to the Astronomical
McCarthy, D. D., Hackman, C., & Nelson, R. A. 2008, AJ, 136, 1906 Almanac, U.S. Naval Observatory (Mill Valley, CA: Univ. Science Books)
McGugan, A. 1967, Abstr. Mtg. Geol. Soc. Am., 145 Simon, J. L., Bretagnon, P., Chapront, J., et al. 1994, A&A, 282, 663
Newcomb, S. 1895a, Astronomical Papers Prepared for the Use of the Sonnet, C. P., & Chan, M. A. 1998, GeoRL, 25, 539
American Ephemeris and Nautical Almanac, Part I: Tables of the Sun, Vol. Sonnet, C. P., Kvale, E. P., Zakharian, A., Chan, M. A., & Demko, T. M. 1996,
6 (Washington, DC: U.S. Govt. Printing Office) Sci, 273, 100
Newcomb, S. 1895b, The Elements of the Four Inner Planets and the Stephenson, F. R., & Morrison, L. V. 1984, RSPTA, A313, 47
Fundamental Constants of Astronomy (Washington, DC: U.S. Govt. Stephenson, F. R., & Morrison, L. V. 1995, Philos. Trans., 351, 165
Printing Office) Vanyo, J. P., & Awramik, W. M. 1985, PreR, 29, 121
NIMA 2000, World Geodetic System 1984, TR8350,2, 3rd edition, Wells, J. W. 1963, Natur, 197, 948
Amendment 1, see pages 3-4 and 3-6 Wells, J. W. 1970, in Paleogeophysics, ed. S. K. Runcorn (London: Academic)
Olah, S. 2009, Gen. Sci. J. Williams, G. E. 1989a, EOS, 70, 40
Pannella, G. 1972a, Ap&SS, 16, 212 Williams, G. E. 1989b, EPISODES, 12, 162
Pannella, G. 1972b, Precambrian Stromatolites as Paleontological Clocks, 24th Williams, G. E. 2000, RvGeo, 38, 37
Int’l. Geol. Congress, Section 1 Precambrian Geology, 50 Williams, J. G., & Dickey, J. O. 2002, 13th International Workshop on Laser
Pannella, G. 1976, Die Naturwissenschaften, 63, 539 Ranging, 4
Pannella, G., MacClintock, C., & Thompson, M. N. 1968, Sci, 162, 792 Zhu, S. X., & Huang, X. G. 2003, Journal of Micropaleontology, 20, 23

12

You might also like