72 



WORK OF THE CARNEGIE AND SUGGESTIONS FOR FUTURE SCIENTIFIC CRUISES 



on the photographic record of the gravity instrument, in 

 order to make for high precision in the corrections and 

 rates of the chronometers used in the determinations of 

 gravity. Provision also was made to obtain the chro- 

 nometer corrections aurally by using coincident beats 

 of a mean-time chronometer with rhythmic radio time 

 signals, or by using the coincident beats of a fast-run- 

 ning chronometer with mean-time radio signals. Whon 

 at sea on board the Carneple the aural method Mras used 

 entirely, for it was not necessary then to swing the 

 pendulums and to develop the photographic trace, and, 

 moreover, it was found that reliable aural comparisons 

 could be made when radio reception was so poor that the 

 signals could not be recorded automatically. The aural 

 method, together with the calculations for the chronom- 

 eter corrections, is described in "The gravity-measur- 

 ing cruiseof the U. S. submarine S-21" (1, pp. 37-38). 



The method is the same as for a coincidence deter- 

 mination between two chronometers; it requires the no- 

 tation of the times indicated by the chronometer and by 

 the radio time signal at the instant of coincidence. Since 

 the method has seemed difficult to some observers, it 

 will be described in more detail. It consists in putting 

 the chronometer circuit, which is alternately open and 

 closed for half-second intervals, in series with the 

 phones through which the radio time signals are to be 

 heard. Since the chronometer rate is altered so that it 

 gains one mean-time second in about sixty-five, the pur- 

 pose of which is to insure several coincidences with the 

 radio time signals during the few minutes of their trans- 

 mission and further to insure that these coincidences do 

 not all occur in the silent intervals which are used to 

 identify the minutes of the signal, the effect is a periodic 

 "appearance" and "disappearance" of the radio signal. 

 Some reflection shows that just when the time signal 

 "disappears" there is a coincidence between the begin- 

 ning of a radio time-signal impulse and one of the chro- 

 nometer beats, and that the "appearance" marks a co- 

 incidence between the end of a particular time-signal 

 impulse and a chronometer break. Since the corrections 

 to time signals which are sent out by the transmitting 

 observatory are for the beginnings of the impulses, it is 

 clear that only the coincidences on "disappearance" 

 should be used. If rhythmic radio time signals are used, 

 then the coincidences may be obtained directly with the 

 beats of a mean-time chronometer. The observer must 

 be prepared at the instant of the coincidence to note both 

 the chronometer time and the radio time. With some 

 practice this can be done easily it the observer counts, 

 from the beginning of an interval in which a coincidence 

 is expected, the beats of the chronometer, which he can 

 always hear in the phones. Suppose a coincidence is 

 expected when the chronometer reads abwut thirty sec- 

 onds. At fifteen seconds, for instance, the observer, by 

 watching the chronometer, notes that beat in the phones 

 which occurs just when the chronometer second-hand 

 moves to fifteen; this beat he calls "fifteen," the next 

 beat he calls "half," the following one "sixteen," and 

 then successively "half," "seventeen," "half," "eight- 

 een," etc., to the instant of coincidence. Thus the ob- 

 server carries the fast-running chronometer time in his 

 head without the necessity of continuously watching this 

 chronometer; a glance at it at any instant tells him 

 whether his counting is correct. To determine the radio 

 signal time at the instant of coincidence, he notes visu- 

 ally the time, to the nearest half-second, indicated by a 

 mean-time chronometer. Then he immediately writes 



down in seconds and half -seconds the time by each 

 chronometer. After this is done he has plenty of time, 

 nearly a minute, to note the number of hours and minutes 

 indicated by each chronometer. To obtain the radio time 

 of the coincidence it is necessary to know, only to the 

 nearest half-second, the correction of the mean-time 

 chronometer on the radio signal. Generally this can be 

 obtained between coincidences by throwing a switchwhich 

 will allow only the radio signals to he heard in the phones; 

 the mean-time chronometer reading to the nearest half- 

 second is then obtained visually at the beginning of a par- 

 ticular minute of the radio signal, or at the middle of the 

 minute if this is identified by the signals. 



The radio apparatus on the Carnegie was located in 

 a cabin on the main deck; the gravity apparatus and the 

 chronometers were below in the main cabin. Leads 

 served to transmit the time signals from the radio re- 

 ceiver through a pair of phones worn by one observer 

 near the chronometers. The circuit was so arranged 

 that as soon as the radio operator had tuned in the time 

 signal he could also hear the coincidences and by means 

 of an auxiliary chronometer note the radio times of co- 

 incidences. These data, together with those obtained by 

 the first observer, provided in effect two independent 

 coincidence determinations. When the time signals were 

 weak, this proved of great assistance and provided a good 

 check. The two sets of coincidence determinations gave 

 corrections to the chronometers which rarely differed by 

 more than about 0.02 second. 



In this connection must be mentioned the importance 

 of obtaining corrections to the chronometers as frequent- 

 ly as possible, in order to minimize the effect of any di- 

 urnal variation in their rates In any case, the daily 

 rates as obtained between time-signal observations are 

 the average rates for the interval between time signals. 

 The duration of a gravity observation is about half an 

 hour. If the actual rates of the chronometers during this 

 half-hour differ from the average rates in the interval 

 between time signals, then an error in the calculated 

 value of gravity results. Data indicating this magnitude 

 of variations in the difference in the daily rates of the 

 tv/o chronometers may be obtained in two ways. One 

 consists in postulating that the difference between the two 

 values of the period of either of the two pendulums in the 

 gravity instrument as obtained from the two chronometers 

 is owing to the difference in their daily rates. The sec- 

 ond method consists of making an accurate comparison 

 of the two chronometers near the beginning of the gravity 

 record and again near the end. This can be accomplished 

 by using the swing of the fictitious pendulum (to be de- 

 scribed later) itself to measure the time interval between 

 adjacent breaks of the two chronometers. The compari- 

 son, thus, is obtained with sufficient accuracy, particular- 

 ly if the two chronometer breaks are not far apart, and 

 further if they are on the same slope of the pendulum 

 curve and near the center of the record as well. 



The difference in the daily rates calculated by these 

 two methods should be in good agreement and thus should 

 provide some check on the determinations of the periods. 

 The deviations between this difference in the daily rates 

 and that obtained from the time signals give some indi- 

 cation of possible fluctuations in chronometer rates. 



Theory of the Apparatus 



The complete theory and description of the appara- 

 tus will be found in "Theory and practice of pendulum- 



