Time
Signals
msf | world time
| uncertainty | basic units | home
When I started
work at the NPL in 1929 the only way of measuring the frequency of an
oscillator was to count the number of cycles between two of the 1-second ticks
transmitted from an observatory. These were primarily intended to give the time
to ships at sea and enable them to determine their longitude, and the
measurement of frequency was a secondary application. The time signals were
accurate to about 0.01s so that it was necessary to extend the measurement over
a period of 24 hours to obtain a value accurate to 1 part in 107 and
this was a great inconvenience. The frequencies of two oscillations can be
compared with a far greater accuracy in a few minutes and experiments were
initiated by the International Scientific Radio Union to find out how such comparisons
were influenced by the propagation conditions, when made at a distance. In one
experiment in 1934 the NPL standard tuning fork was used to modulate the
powerful 193kHz transmitter at Droitwich. It was received and measured at a
number of standards’ laboratories in Europe and the results agreed to 1 part in
107, representing the highest accuracy achieved in international
comparisons at that time.
During this period
I amused myself by trying to measure the weak transmission of 5MHz made from
Washington USA for their own local standardisation purposes. It could be
received for a few hours during the night and I compared its frequency with
that of the 250th harmonic of our 20kHz quartz standard. Although
the accuracy of measurement was good to 2 parts in 108 and the
standards in the two countries were believed to be known and stable to the same
degree of accuracy, erratic variations of several parts in 107 were
observed. These were attributed to Doppler effect changes caused by variations
in the height of the reflecting layer. On occasions a regular beat was obtained
but still gave a frequency difference of as much as 1 part in 107
due to a seemingly steady change in the height of the layer. The results showed
that on average the frequencies could be compared to 2 parts in 108an
that a detail study of the results could give useful information concerning the
height of the reflecting layers.
MSF
All standard frequency
transmissions from the UK were suspended during 1939-1945 but a new service
with the call sign MSF was inaugurated in 1950. It was too expensive a service
to be met from the NPL budget but Sir Edward Appleton, who was then secretary
of the government’s Department of Scientific and Industrial Research (DSIR),
realised their importance, having himself used the early NPL transmissions in
his fundamental work on the study of reflections at the ionosphere, made the
necessary financial arrangements. The transmissions were operated by the PO
(Post Office) at Rugby on behalf of the NPL which supplied the frequency
standard. The frequencies used were those agreed internationally at 2.5, 5 and
10MHz but we added a frequency of 60kHz to cover Europe without ionospheric
reflections. On one of my visits to the USA I was introduced to Jack Pierce at
Harvard. He was investigating the application of frequency transmissions to air
and sea navigation and I suggested that he might like to try and receive our
frequency-controlled transmission at 60kHz. He found that he could receive it
for a few hours at night and he was so impressed by its stability that he put
in a formal request that the powerful 16kHz transmitter, which was also sited
at Rugby, should be controlled by the same standard. The PO agreed to do this
and this station, having an almost world wide coverage without ionospheric
reflections, was the first to be controlled by the national frequency standard.
When it was later controlled by our atomic clock it made atomic time available
throughout the world.
World wide
synchronised time
The close co-operation with the USA made it
easy to establish a world wide synchronised time service. As atomic clocks
became available in the USA, Wm. Markowitz used them to control the service
from the US Naval Observatory. The value of the atomic unit had been agreed but
atomic clocks do not, of course, give the time of day and, although this is not
required with the same accuracy, it is useful to have agreement throughout the
world. To establish the accuracy of achieving this, time signals were
interchanged between Washington and the NPL and a small difference of 2
milliseconds was revealed and resolved by mutual agreement between us including
the Royal Greenwich Observatory. As other services came into use with atomic
standards they adopted the same time, and an international synchronised time
service was established without any formal discussions. More precise time
comparisons were made later by re-transmitting the signals from the satellite
Telstar, and avoiding errors due to ionospheric reflection. In this way
synchronisation to 1 microsecond was achieved.
Uncertainty
It was assumed in
all these comparisons that the transmit time was the same in both directions,
which could be proved only within the limits of experimental accuracy. This
residual uncertainty was a great worry to Einstein, and the basis of his
theory. But it was only an example of the uncertainty present in all scientific
results. It is usually assumed that in order to obtain a consistent structure
of science the units of measurement must be independent, and every effort was
made to secure this. As measurements became more precise it should have been
realised but was overlooked that this independence cannot be fully achieved
because the units are determined in the complex earth environment in which all
units are involved. The distinguishing feature of science is that experiments
are repeatable and that the limits of error can often be made very small, and
are, themselves, determined by experiment. The limits in time and frequency
measurement are by far the smallest of all.
Basic units
I was fortunate to
take part in a transformation of two of the basic measurements in physics; time
from the solar system and astronomy to the atom and physics, and length from a
metal bar to the velocity of light multiplied by time. In both cases there was
an enormous increase in accuracy and in the case of time measurements an
enormous simplification as a bonus.
