Located close to the Thames in London, Greenwich was particularly vulnerable to bombing raids when World War 2 began. As a result, two emergency time stations were set up. The first (known for security reasons as Station A) was at the Observatory’s outstation, the Magnetic Observatory in Abinger. The second (Station B) was at the Royal Observatory Edinburgh.
The following extracts relating to the Greenwich Time Service are taken from:
They should be read in conjunction with:
In the autumn of 1940, as a result of representations made by the Astronomer Royal, it was decided that the Observatory should be equipped for the purpose of duplicating the various time signals normally originating from Greenwich. (Provision had previously been made for transmitting the B.B.C. “six-pips” signals only.) The work of installation was to a large extent carried through in 1940 December, the equipment sent from Greenwich including the Free Pendulum Clocks Shortt Nos. 11 and 16. By the end of December the work of installation had fortunately been carried to a point which allowed the transmission of rhythmic signals to Rugby, and as a result the Observatory was able to operate the Rugby signals, although under difficult conditions, in 1941 January when, in consequence of enemy action, it was impossible to operate this service from the Greenwich station at Abinger. At the same time the work of installation proceeded. From 1941 May to 1943 March 29 and from 1943 October 27 to the present date the rhythmic signals radiated from Rugby at 10h U.T. were normally operated from Edinburgh, but originated from Abinger on a few occasions when, owing to technical faults, transmissions from Edinburgh were not possible or advisable. A large number of the Rugby transmissions at 18h U.T. and of the transmissions from British short-wave stations (which were commenced in 1942 July) have also been operated from Edinburgh. On a few occasions the Observatory has been called upon to operate the B.B.C. “six-pips” signals. Hourly signals to the Post Office have been supplied and have been available for the control of the “speaking clock”.
In 1943 February a direct line link was installed between Abinger and Edinburgh whereby the clocks at each station were made available at the other. Early in 1944 a quartz clock of the Post Office Group IV type, the first of a set of such clocks constructed for the Royal Observatory, Greenwich, by the Post Office Radio Branch Laboratories, was installed at Edinburgh, and commencing 1944 February 26 the Rugby signals both at 10h and 18h U.T. (at the time the 18h. signals were being operated from Edinburgh) originated from this clock, the phonic motor of which was constructed so as to provide for the transmission of rhythmic vernier signals. A similar clock was installed at Abinger shortly afterwards (together with another Group IV clock, the phonic motor of which delivers mean time second impulses).
Throughout the whole period transit observations were secured at Edinburgh on all possible occasions, and were combined with observations obtained at Abinger for the purpose of deriving clock errors.
Considerable developments have been made in the general technique of the Time Service during these war years. In the following paragraphs reference will be made only to development work carried out at Edinburgh with the object of improving the service. This report is thus to be regarded as supplementary to the Greenwich reports, to which reference must be made for a complete account of the various changes that have been made in the pre-war technique. In addition to the requirement of providing signals giving absolute time to an accuracy adequate for navigational and surveying purposes, the signals radiated from Rugby at 10h U.T. are now intended to provide a standard of frequency, the desideratum being that consecutive 10h signals should define an interval of 24 hours with an accuracy of the order of a millisecond. The development work which has been carried out at the Edinburgh end has been concentrated on this aspect of the service.
Throughout the years 1941 to 1943 the operation of the service was based on Shortt free pendulum clocks. During this period the performance was improved. The original circuit arrangements were modified, and in addition to the development (to which reference is made below) of line transmission, radio reception and chronographic measurement, a technique was developed for using a “mean clock” for signal transmission purposes, the effect of clock erratics being reduced by averaging as many clocks as were available. The process developed was similar in principle to the process previously developed at Greenwich for the purpose of basing definitive corrections to the signals on a “mean clock”, but there were differences in detail which arose from the circumstance that for signal transmission purposes clock errors have to be extrapolated. During 1942 a fundamental change was made in that the service was ultimately based on quartz controlled clocks situated at other institutions, and the process of extrapolation of the errors of the pendulum clocks at the Observatory was reduced to extrapolations, for periods of 24 hours only, of, the relative errors of the pendulum clocks and the quartz controlled standards, these relative errors being obtained via a radio link which employed the 10h Rugby signals themselves. It had become apparent at this stage that the performance of quartz controlled clocks was of an altogether higher order than that of the best pendulum clocks. During 1942 and 1943 some investigations were carried out on the erratics of free-pendulum and quartz clocks, the results being published in a paper by Greaves and Symms (M.N., 103, 196—the free-pendulum clocks designated X, Y and Z in this paper are respectively Shortt 4, Shortt 11 and Shortt 16). The conclusion reached that the P.O. type quartz controlled clocks are very superior in performance has been more than confirmed by subsequent experience. As regards their erratics of rate (designated in the paper by types B and C) the mean absolute scatter in daily rates has been found (from an analysis, made at the Observatory, of the results of beat frequency comparisons supplied by the Post Office Radio Branch Laboratories) to be of the order of ±0-05 milliseconds per day. This figure represents a combination of erratics of types B and C, and the performance indicated thereby contrasts strikingly with that of the Shortt clocks for which the mean absolute scatter of daily rate due to type C erratics alone seems to be of the order of ± 1 to ±2 milliseconds per day. The “wanderings” of pendulum clocks are in fact much greater than those of quartz controlled clocks of the Post Office Group IV type. As regards erratics of type A, an investigation, using an electronic chronometer, at the Post Office Radio Branch Laboratories has shown that the phonic motors of the quartz clocks introduce random errors with a standard deviation of the order of ± 0.1 milliseconds. The Observatory is indebted to the Radio Branch, Engineer-in-Chief’s Office, General Post Office, for permission to quote this result and the result of the analysis of the beat frequency comparisons.
It was clear that the state of affairs in which the signals were operated by pendulum clocks was unsatisfactory, even though the service was ultimately based on quartz controlled clocks. But in some respects the experience gained with pendulum clocks was of permanent value. The performance of the “mean” pendulum clock was good enough to justify efforts being made to reduce the errors of radio reception and chronographic measurement and to secure a greater degree of constancy of the transmission time over the line system connecting the Observatory to Rugby. The technique of radio reception and of chronographic measurement were developed and, largely owing to the very valuable co-operation given by the Post Office Engineering Department (Scottish Region), considerable reductions were effected in the errors introduced by variable line lag. The result was that when the quartz clock was installed in 1944 it was possible to take advantage of its superiority and to reduce materially the errors of individual 24-hour intervals defined by consecutive G.B.R. signals at 10h. But it became clear that the full superiority of the clock was not being utilised. More improvements were made in the chronographic technique, and in the later months of 1944 efforts were successfully made to secure a further reduction of errors due to line transmission. A series of experiments, which included oscillographic investigations of build-up, were carried out with a view to improving the method whereby the signals are fed on to the line, and the previous circuit arrangements were modified. The Post Office contributed to this by supplying relays of a greatly improved type, and they took the important step of discarding the arrangement whereby a traffic fine channel was switched over to signal use for a short period only, and provided a permanent channel, with a loop at Rugby feeding a return channel, and available for test at any time. This channel became available in 1945 January; an arrangement was made whereby the stability of the line system was kept under observation at the Observatory, and on receipt of a message that re-adjustment of the system was necessary the Post Office engineering staffs have promptly taken the necessary steps. In addition they have themselves made regular tests of line stability, employing the same method of testing as has been applied at the Observatory. This method is an adaptation of the “chaser” method devised at Greenwich for measuring the total lag of a looped line. In the adaptation a meter reading indicates whether the line system is in good adjustment.
In 1945 March arrangements were made whereby a test signal is sent out from the Observatory and radiated by Rugby at 9h 30m U.T. on each day, a direct measure of lag being thus made half an hour prior to the signal at io*1 U.T. As a result of the improved stability of the line itself (mainly due to the stringent control), of the improved method of feeding signals on to it, and of the provision for direct measurement of the lag at 9h 30m U.T., the standard deviation of the measured errors of individual intervals of 24 hours has been reduced to ±o.6 milliseconds (the mean absolute scatter being ±0.5). This figure represents errors which are additive to the errors in the 24-hour intervals introduced by the general trend of the signals. It is the practice to adopt a parabolic arc on the plot of clock errors against date, and to base the signal transmissions on this adoption. The adopted curve is revised from time to time as further transit observations become available, attention being paid to the comparisons of the signal transmitting clock with other quartz standards. The revisions are not made daily, but a single adoption is used for a period which may extend over several weeks, and consequently the resulting errors in signal rate persist at the same value over extended periods. These are the errors of general trend. From a preliminary discussion covering the period 1944 August 1 to 1945 April 17 their mean absolute value was ±0.6 milliseconds per day with extreme values of +1.5 and -2.3. The extremes are larger than one would like, and are due partly to the limitations of the present method of observing transits and partly to the circumstance that the Observatory clock is mains operated, the crystal frequency being liable to disturbance from surges and failures of mains voltage. There is no doubt that these errors of general trend could be diminished by the provision of better types of transit instrument, and by the provision of battery drive for the quartz clock. As regards the additional errors in individual intervals of 24 hours, part of the measured mean absolute scatter of ±0.5 milliseconds quoted above is certainly due to errors of radio reception and of chronographic measurement, and the real signal performance must be better than is indicated by this figure. The precautions which have been taken include the reduction of the errors of radio reception by oscillographic control of the tuning and of the output voltage of the receiver, and the technique of chronographic measurement has been refined as far as possible. It is felt that there is little chance of reducing the errors of reception and measurement still further as long as classical chronographic methods are employed. In fact, as has been indicated, some considerable effort was expended in the achievement of the standard of accuracy that has actually been reached, and the same remark applies to recent signal performance which has been the result of close co-operation between the Observatory and the Post Office Engineering Department (Scottish Region). The reduction of the errors introduced by line transmission to the level indicated by the measured mean absolute scatter of ±0.5 milliseconds for individual 24-hour intervals is regarded as very satisfactory.
It is now expected that the participation of the Observatory in the Greenwich Time Service will soon be terminated, and that a resumption of pre-war activities will be possible.
The participation of the Observatory in the Greenwich Time Service was terminated at the end of 1946 January. The signals radiated from Rugby at 10h U.T. originated from the quartz clock A1 at Edinburgh up to 1945 December 22, after which date these signals were operated from Abinger. The accuracy of signal transmission referred to in the previous report was maintained, and, in fact, somewhat improved, the mean absolute measured departures of the 10h signals from the times by the clock A1 at which it was intended to transmit them being ±0.25, ±0-.24 and ±0.25 milliseconds for the months of October, November and December respectively. These figures include relay errors of feeding the signal on to the line to Rugby, errors of line transmission, errors of radio transmission at Rugby and errors of radio reception and chronographic measurement at Edinburgh. The corresponding contribution to the mean absolute error of individual 24-hour intervals defined by successive signals at 10h U.T. would be √2 times the above figures, giving a mean of ±0.35 milliseconds as against the value ±0-5 milliseconds quoted in the previous report. To these errors in individual 24-hour intervals must be added the errors of general trend for which the mean absolute value quoted in the previous report was ±0.6 milliseconds.
During the latter half of 1945 it was noticed that the clock A1 at Edinburgh and the three quartz clocks at the Post Office Radio Branch Laboratories at Dollis Hill all appeared to change their rates, as determined from transit observations, at Edinburgh and Abinger, by about the same amount (about 4 milliseconds per day) at about the same time (the middle of September). This is obviously a matter requiring careful investigation, and is receiving such at Greenwich. It is understood that the same phenomenon has been noticed at other observatories.
An investigation of the errors which affect determinations of time with the Edinburgh Transit Circle has been carried out by Mr. Symms (whose services were on loan from the Royal Observatory, Greenwich) and Miss Penny [1946MNRAS.106..390S]. They find a marked correlation between the individual clock errors determined on various nights and the measures of collimation error. This shows that erroneous measures of collimation contribute materially to the total probable error of a determination of time. The total probable error of a single night’s time determination (using the mean of ten stars) with the Transit Circle is ±20 milliseconds when personal equations are applied to the results from the several observers. But that part of the whole probable error not due to errors in measuring collimation and level is ±10 milliseconds or less, this figure including the contribution due to erroneous star places. The impersonal micrometer attached to the Transit Circle is hand driven.
The Time Service equipment was dismantled and returned to Greenwich during February.