On 2017-10-15 03:28 PM, Steve Allen wrote:
On Sun 2017-10-15T18:59:31+0000 Michael.Deckers via LEAPSECS hath writ:
Thank you for these interesting primary sources!
The more I reflect on this the more I realize that Nicolas Stoyko
attended and took notes on many of the astronomy/time/radio meetings
during the 1950s and 1960s. Lots of the history of who said what when
is in the opening sections of issues of Bulletin Horaire, and detailed
in a way that is not typically found in the official proceedings.
Yes. These references are valuable for the insight they give to the
thinking at the time. Thanks.
The 1955 meetings are interesting because atomic time was a thing,
but atomic timescales were not yet a thing anyone had done, and
Ephemeris Time was a thing that a few people had done.
So in all those discussions it is clear that Universal Time is
unquestionably understood to be a measure of earth rotation.
I agree.
The slowing of Earth rotation was suspected early, including by Simon
Newcomb They did not have confidence in the reasons for this, and
certainly not accurate enough observation nor constant frequency to
include it in the Tables of the Sun.
On the possible variability of the Earth's axial rotation, as
investigated by Mr. Glasenapp- Simon Newcomb
http://www.ajsonline.org/content/s3-8/45/161.extract#
The references you cite suggest they still expected UT2 could somehow be
made to be a uniform scale. Not until atomic time and timescales are
established does it become undeniable the Earth rotation was slowing.
It seems the critical development was Essen and Markowitz's
determination of the length of the atomic second with respect to
Ephemeris Time, completed in 1958. Those efforts are explained in detail
by Leschiutta:
The definition of the ‘atomic’ second, Sigfrido Leschiutta
http://geodesy.unr.edu/hanspeterplag/library/geodesy/time/met5_3_S03.pdf
The numbers they'd found were to "stick", resulting in the definition of
the SI atomic second and leading to the atomic timescales, eventually to
TAI.
Through the early 1960s the many atomic clocks were "coordinated",
giving rise to coordinated atomic timescales and the small sub-second
frequency and step adjustments required to keep atomic time is step with
solar time, as we see in the historical Leap Seconds record.
I find this article by L. Essen most informative of the situation in
timescales in 1968, and gives a good summary of how they'd gotten there:
Time Scales, L. ESSEN
Metrologia, Vol.4 , No.4, 1968
http://www.leapsecond.com/history/1968-Metrologia-v4-n4-Essen.pdf
The predisposition toward a single disseminated atomic timescale seems
to have taken hold, as he states:
"It would of course be possible to use separate and independent scales
of atomic and astronomical time but this possibility already seems to
have been rejected, and rightly so in my view, since it would lead to
confusion and duplication of effort."
And he concludes with his suggestion for 1s adjustments, which would
emerge as "Leap Seconds":
"If it is necessary the minute markers should be moved by 1s at a
prearranged date if the deviation between AT and UT2 exceeds 0.5s."
In 1971 J. McA. Steele gives a good explanation of Leap Seconds as they
were to be implemented, including an appendix showing the CCIR approved
Recommendation 460.
THE IMPROVED UTC SYSTEM TO BE INTRODUCED ON I JANUARY 1972
by J. McA. Steele, August 1971,
https://journals.lib.unb.ca/index.php/ihr/article/viewFile/23857/27642
Barnes gives an excellent summary of the timescales in 1973:
BASIC CONCEPTS OF PRECISE TIME AND FREQUENCY, Barnes, 1973
http://tf.boulder.nist.gov/general/pdf/1593.pdf
I observe -
The development of atomic clocks and timescales had confirmed a new
physical phenomenon; constant frequency time by atomic science. The new
systems increased accuracy by at least three orders of magnitude over
earlier clocks. This, together with improve astronomy, made it possible
for the first time to quantify the discrepancies between uniform
constant frequency and astronomical time, clearly revealing the Earth's
unpredictably slowing rotation.
Thus there are two physical phenomenon, atomic time, and astronomical
time, obviously demanding two timescales. But two timescales was seen as
difficult and there was a question of priority; which was to be the
*reference*? It could really have been done either way; atomic time
defining "seconds" (as TAI has become) with astronomical time defined
with respect to atomic time (as UTC and UT1 have become), or other way
round; astronomical time defining "seconds" ("solar seconds"?) with
atomic time defined with respect to astronomical time. Atomic time won
that discussion.
UTC is obviously and intentionally an approximation of mean solar time,
a compromise between the uninterrupted incrementing TAI SI seconds
counting method and the Gregorian YMDhms counting method modified with
the Leap Second (23:59:60) to define the boundaries between days. This
has left us with the incommensurability between UTC and the classical
Gregorian calendar, local time-of-day (conventional clocks), and
computer/electronic implementations of Gregorian, local time, and
time-of-day.
Speaking of meetings, ITU-R WP7A meets in 10 days and looks likely to
discuss leap seconds.
There continues to be debate and focus on retaining a single timescale
and yet this was never actually the case. UTC is really two timescales,
(an abstract form of) TAI, together with an approximation of mean solar
time via the Leap Seconds modification of the Gregorian counting
algorithm. Further, Rec 460 also specifies 1/10th second resolution
DUT1, and the radio time signals disseminate this signal too. So, that's
really *three* timescales.
Proposed solutions to the UTC v.s. "computer time" problem have focused
on delaying the adjustments or, most prominently, simply eliminating
Leap Seconds. I think these sorts of solutions will meet the same fate
they have in the past, that is, "no action", because it is difficult or
impossible to substantially change any standard or protocol that has
gained wide acceptance, and UTC might be the most widely used standard
of all time. The exception might be the pure Gregorian calendar itself,
and therein lies the difficulty; two very widely used timekeeping
standards that are incommensurate.
The "smeared" time dissemination services (Google smear, etc) are used
to eliminate the Leap Second from the receiver's view of the timescale,
restoring a 86400-second-day to the classical Gregorian algorithm. Those
methods focus on smearing the 24 or 12 hour periods surrounding the Leap
Second. Unfortunately they do not match one another and introduce a
rather extreme frequency shift between TAI/UTC and the resulting
"Leap-Second-free" timescales on Leap Second days.
I feel UT1 does not receive the attention it deserves. In this context
UT1 might be used to fix UTC v.s. "computer time" conundrum by supplying
the information to more faithfully and gradually distribute the
"smeared" frequency shifts between successive Leap Seconds, retaining
traceability to UTC. I think the IERS already produces the raw
information in Bulletin D (DUT1) on which an approach like that could be
based.
I hope the participants of ITU-R WP7A and contributing organizations
might consider redirecting their efforts from seeking to eliminate Leap
Seconds to a more robust technical approach that recognizes the need for
more than one timescale.
-Brooks
--
Steve Allen<[email protected]> WGS-84 (GPS)
UCO/Lick Observatory--ISB 260 Natural Sciences II, Room 165 Lat +36.99855
1156 High Street Voice: +1 831 459 3046 Lng -122.06015
Santa Cruz, CA 95064http://www.ucolick.org/~sla/ Hgt +250 m
_______________________________________________
LEAPSECS mailing list
[email protected]
https://pairlist6.pair.net/mailman/listinfo/leapsecs
_______________________________________________
LEAPSECS mailing list
[email protected]
https://pairlist6.pair.net/mailman/listinfo/leapsecs
