Journal of the Society of Telegraph Engineers, 1873, pages 433-441:

ON  THE  ALLEGHENY  SYSTEM  OF  ELECTRIC  TIME  SIGNALS.

By  Prof.  S.  P.  LANGLEY.

THE necessity of a uniform standard of time for the railways of the United States is one which is growing into importance with the increasing extent of our railway system, and we are, ere long, in this country, to be called on to settle for ourselves a practical problem which has been already solved in England, and which is beginning to make its demand for solution upon the managers of our railroads.
    Although the introduction of the plan in this country has been comparatively recent, the number of American observatories which thus distribute time is so considerable that the most partial account of their methods, and the extent of their work, would exceed the limits of such an article as the present. In this, the only arrangements described are those in use at the Allegheny Observatory, with which the writer has become familiar from the responsibility of their initiation and superintendence. It is proper to add that, were he writing a history of the progress of electric time signals in the United States, other observatories which have before employed not dissimilar means, would receive earlier mention, and that his own part in introducing these signals at the Allegheny Observatory has been less the contribution of any novel device than an adaptation of what seemed the best features of plans in use abroad, their arrangement in a form adapted to the needs of American railways; and the supervision of their application to the wants of cities and individuals. In doing this a great number of ingenious devices have been examined, and if the system to be described appears to be one of the simplest, it has yet been reached only after much care in setting aside all which would not bear the test of practical trial.
    The subject was first specially considered at the Allegheny Observatory some three years since, and a plan was arranged for the managers of the Pennsylvania Central Railroad in 1869. Previously to this, however, at the request of some jewellers of Pittsburg, the time had been transmitted to their stores, at a distance of some miles from the observatory. The system now described has been in use for nearly three years, in furnishing the Pennsylvania Central Railroad with its official standard of time, and by it the time is now sent daily to Philadelphia on the east, as far as Lake Erie on the north, and to Chicago on the west--regulating the clocks on a number of minor roads over whose wires it goes, as well as on those of the principal southern lines connecting the Atlantic with the Mississippi. Thus passing, as it does, over several thousand miles daily, it is believed to be at present one of the most extended systems of time distribution in the world.
    The observatory is on the summit of the ascent, on the northern side of the valley of the Ohio, about two miles in a direct line from the offices of the Western Union Telegraph Company in Pittsburg, and rather more from those of the Pennsylvania Central, and Pittsburg, Fort Wayne, and Chicago roads. It is connected with these points by three independent lines of telegraph. One of these runs to the Western Union offices, and thence to the stores of a considerable number of jewellers in Pittsburg. This is called the "jewellers' line." The second, connecting the observatory through the offices mentioned with eastern Pennsylvania and New Jersey railways, and also with Chicago, is known as the "railroad line." The third, consisting of a double wire or "loop," communicating with the city, is employed occasionally for the observatory's own messages, and when (as, for instance, in longitude determinations) it is wished to send sidereal time, without interrupting the regular transmission of signals from the mean time clock. In the transit room, in the western wing of the observatory, are kept the sidereal clock, by Frodsham, of London, and the principal mean time clock, by Howard, of Boston.
    On the escape wheel arbour of this, the standard mean time clock, and turning with it once a minute, is a wheel cut with sixty sharp radial teeth, of which those corresponding to the 50th, 51st, 52nd, 53rd, 54th and 59th seconds of the minute have been removed by a file. Near the clock is a "repeater," the circuit through whose coils passes through a local battery, through a second clock in the computing room, and then through the standard clock. Each wire terminates in a delicate spring close by the wheel just mentioned. While the extremities of these springs, which are shod with gold and platinum, rest in contact, the circuit is unbroken; it is opened by the minutest lifting of one from the other, and this is effected automatically by means of a ruby attached to one of them, and placed within reach of the wheel above mentioned. As each of these teeth passes, the ruby, and with it the spring, is lifted through a minute distance. (Not in practice more than one one hundreth of an inch, and usually much less.) Once a second, therefore, the circuit is opened during a period of probably less than a twentieth of a second, and as the wheel advances a tooth with each vibration of the pendulum, the armature of the repeater is raised each second of the minute until the 49th is completed.
    Since the teeth corresponding to the next five seconds have been filed away, during these seconds the jewel is not touched nor the circuit opened. The consequent silence of the "repeater's" beats draws attention to the fact that the end of the minute is approaching, its completion being indicated by the short pause caused by the absence of a tooth at the 59th second.
    This action is repeated in every minute of the twenty-four hours without variation. The particular second is thus identified, but one minute is (so far as the action of the standard clock is concerned) not distinguished from another. To do this is the work of the subsidiary clock in the computing room, through which the local wires are led, as has been mentioned. The subsidiary clock (made by Howard, of Boston) may be called for distinction the "journeyman," and its principal office is not to give the time but to interrupt the circuit, which it does on or near the completion of the 58th minute, closing it again about half a minute before the completion of the hour. When the circuit is opened by the journeyman the repeater is silent for a minute and a half; when it is closed, the standard is again heard ticking on the repeater, and the ensuing short pause evidently precedes the first second of the first minute of the hour. The time is thus wholly derived from the standard clock, and is independent of any other; the journeyman having no power to control or in any way re-act upon the primary, and being only able to interrupt the messages it sends, not to falsify them.
    The mechanism for effecting the transmission of the time is essentially that already described, but more is needed to insure against possible interruption. This may occur from several causes, prominently from oxidation of the platinum or gold contact surfaces, when the current must be interrupted while they are cleaned, if there be no other clock. To meet this contingency a chronometer of peculiar construction was made for the observatory by Frodsham. It resembles the ordinary marine chronometer in external appearance, but contains in miniature the apparatus for breaking circuit already described, the wheels being cut so as to give the same signal of the approaching end of the minute as the standard clock. The peculiarity consists less in this, however, than in a device by means of which it can be caused to gain or lose any fractional part of a second, or any number of seconds, without being stopped, and without any disturbance of its normal rate, except while the change is being effected. This chronometer is to replace the prime clock in the circuit, during any temporary stoppage of the latter for repair or adjustment.
    The mechanism which has just been described acts in connection with the local circuits of the observatory--one battery being employed for the sidereal clock and chronograph, and another for the mean time standard. Any interruption of the main external circuits is shown at once by the action of a galvanometer in each, which makes an audible and visible signal when the circuit is opened. The accessory apparatus, such as batteries, relays, switchboards, and so forth, which are used in every telegraph office, it will be superfluous to describe here in detail, but before following the operation of the electric current, outside the observatory, it will be well to speak of the method which has been adopted as likely to ensure most accuracy in the time keepers which control it.
    The transit instrument in the western wing is of four inches aperture, and with it and the chronograph, observations for time are made on every fair night of the year except on Sunday, when, if complete determinations have been made on the preceding night, none are taken. The instrument is of sufficient power to follow the principal nautical almanac stars in the day, and these are used (or more rarely the sun) when the weather permits if the usual night observations have been missed. From three to six stars are customarily taken, the azimuthal error of the instrument being found from the observations of each night, after the other corrections are applied, and the results determined from the chronograph and the sidereal clock. The mean error in the resulting determination of the sidereal clock correction is from three to four hundredths of a second, but it cannot be assumed that that of the mean time standard is known within these limits, except at the time that the observations are freshly made.
    It may be desirable to point out where the system pursued here differs from that in which a few signals are sent at stated hours, as at Greenwich. In the case of the time ball, for instance, dropped daily by a clock at Greenwich, mean noon, it is customary to compare the mean time clock which drops it with the sidereal time a few minutes before twelve. If it (the operating clock) be slow it is caused to gain, and if fast, caused to lose an amount needed to bring it to coincidence before the automatic action gives the signal.
    The time of this signal is nominally exact, but in fact involves the variations in rate of the standard clock or clocks which are treated in the comparison as having their errors absolutely known. It is by no means meant to criticise this procedure, but to point out that an error must exist where the rates of the clocks are treated as constant intervals between observation, no less real accuracy is reached in the method employed here, in which (as the signals are being constantly sent) the signaling clock has no less nominal error at noon (for instance) than at any other hour.
    When the sidereal clock has entered its beats upon the chronograph, during the time of observation, the record is not interrupted until, the mean time standard having been put into the same circuit, both clocks have automatically entered their time on the sheet together, and the break-circuit chronometer has done so also. The sheet being removed, and the breaks of the transit observer measured, the comparison of the various clocks with electric attachments are taken by measurement on the same sheet, and the others compared with the sidereal clock by noting coincidence of beats by ear. The resulting errors of all are then determined, reduced to a common epoch, and entered in a permanent record kept for the purpose in the following form:
    (ΔT, δt, being the usual symbols for the respective corrections of error and date):
    Aug. 10, 1872.     Time stars { η Herculis,
α Camelop,
χ Ophinchi,
δ Herculis,		 }     A. E. F.,
            observer.

 
    At mean 9h       ΔT.        δt.
Sidereal clock, 7s. 32+1s. 18
Break-circuit chron.+2m. 22s. 18+3s. 30
Cron. 3242,+50s. 05+3s. 11
Mean time standard--00s. 27+0· 46

    The mean time clock is here 0 27 fast by actual observation, but when the next comparison is made the following morning (at 21 hours) its error can usually be obtained only by comparison with another clock. If it be compared with each of the other clocks in turn, each, owing to the variations of its rate during the night, will probably give a slightly different, result--but supposing all the time keepers equally reliable, the probable error will be less, in taking the mean of the four, than by any single one.
    The above corrections for error and rate having been applied to the sidereal clock, a comparison is taken with it in the morning, and the resulting time of the mean time clock noted, on the assumption that the sidereal clock is an exact standard. The same comparison is made with each, after the respective corrections and rates have been applied, each being successively treated as an independent standard.
    The results will then be entered in this form:
          
     1872.   August 10d 21h
Error of mean time standard,--0s. 19 (by sidereal clock).
" " ""0s. 05   "  break-cir. chron.
" " ""0s. 11   "  chron. 3242.
" " ""0s. 04   "  its own rate.

The mean or "adopted" error of the mean time standard is then--
 
        -- 0s. 17
________ = -- 0s. 04
4

    In the absence of anymore absolute criterion the time of the standard in this instance is assumed to be kept four one-hundredths of a second fast, and this value is adopted and treated as though it represented an error determined by direct comparison with the stars. The clock will be compared again at 9 in the evening, and when this "adopted error" exceeds 0·25 such a change is made in the pendulum as will correct the error--not abruptly, but gradually during the ensuing twelve hours.
    It is of course impracticable to stop the clock and raise or lower the adjusting screw twice daily for such minute corrections, and many ingenious devices have been proposed for effecting the change without stopping the instrument. One of these, as applied to a chronometer, has already been referred to; another (employed at Greenwich) involves the use of a small bar magnet permanently attached to the pendulum, and swinging with it; and still another the changing tension of a long spiral spring, which connects the "bob" with the clock case.
    After considering many such plans, that adopted was the old one, familiar to most observers, of placing weights on the top of the bob of the pendulum, and then adjusting the bob by the screw till it runs with them approximately, after which a small increment or decrement of the weights will keep the clock under control. This plan has the advantage of employing as an agent gravity, whose effects can be reckoned on with more certainty than electricity or the tension of a spring. In common with the others it has however, as commonly employed, the defect that when changes are made daily or oftener the rate of the clock cannot be ascertained, and that reliance must be placed at the times of comparison only on other clocks whose rates are undisturbed. The writer has, therefore, found it advantageous to use these weights quantitatively, by making them of a size such as to cause a gain of one second a day; 01 an hour, etc. Weights representing three or four seconds are kept on the top of the bob, so that their removal will retard the clock, if desired, to that amount.
    A record is kept in which the comparisons in the tabular form above given are entered, twice daily, the amount of the weights and the consequent rate which the clock so controlled would have had with an undisturbed pendulum being noted likewise.
    The barometer and clock case thermometer are also read twice daily, for the purpose of laying down curves representing the separate effects of temperature and pressure. Another curve, whose ordinates represent the algebraic sum of the corresponding ordinates of the first two, shows the combined results of both, for comparison with still another representing the clock rates. These are chiefly useful in the occasionally long intervals of cloudy weather which occur in winter. At such times the clock rates are obtained by interpolation from the curves, and "weighted" according to the degree of dependence on each clock before making up the final or "adopted error" of the standard. When observations are obtained daily, however, such precaution is needless.
    Those who are aware of the number of patented devices for controlling distant clocks by electricity, may perhaps feel surprised that so little mention has here been made of their use.
    Some of these are of extreme ingenuity and much promise, and the English patents covering such points are alone to be reckoned by scores. Plans have been submitted to the writer by which the clocks along any number of miles of road could be set right, and brought to uniform time in a few seconds, by the operator at the observatory, and these plans appear feasible. The arrangements adopted here, as the reader will observe, do not greatly differ from these employed in telegraphic determinations of longitude, and in fact a prolonged examination of very many ingenious devices for directly controlling distant clocks led the writer to set them all aside, and to employ methods not differing in principle from those in use already, for purely scientific ends, in most American observatories.
    Of the very numerous plans for controlling distant clocks that of Jones (now well known) appears to be the best, but even this is not quite reliable where the circuit is a long one. The clocks described have subsidiary apparatus enabling them to send controlling currents on the Jones plan, but thus far its use has been confined to the observatory, has therefore been hitherto done by the sending of signals, through which distant clocks may be regulated, but without employing means for their control, and though this is done over a very extended field, a brief description of it, under the three divisions into which it naturally falls, will suffice
    1st. The supply of time to watchmakers and jewelers. The "jewelers wire" passes through the Western Union telegraph offices and the stores of the principal jewelers of Pittsburg. Beside each "regulator" is a telegraphic sounder, on which the observatory time is heard constantly ticking, and by which almost, if not quite all the clocks and watches of the city are thus at second-hand regulated. There is, in this uniform and recognized standard, everywhere accessible, a convenience to watchmakers, of course, but still more to the public, as the discrepancies between clocks, public or private, which cause so many lost minutes in the day to each person in a city, that their aggregate represents a large draft upon the time of the business public, disappear.
    Applications have been received from watchmakers in neighbouring cities, and at a considerable distance from Pittsburg, for this telegraphic supply of time, which it has not always been possible to accommodate, but which have been welcome, as showing a public appreciation of the utility of the work.
    2nd. The supply of time to railroads. The watchmakers and jewelers are in permanent telegraphic connection with the observatory by a wire which is devoted to their use--but distant cities, such as Chicago or Philadelphia, can be reached only by the wires of the telegraph or railroad companies which are too valuable to be exclusively employed for this purpose. The method used on the Pennsylvania Central, and Pittsburg, Fort Wayne and Chicago roads, will sufficiently illustrate the system as applied to railways.
    A special wire connects the observatory with the office in which the wires owned by these roads unite. In this office is a small bell, which is struck lightly every second, in the manner described, and except for the pauses to designate the minute and hour, continues to sound unintermittingly, affording to the conductors and other employés specially concerned in the time a means of ready comparison, even without entering the building.
    At 9 and at 4, Altoona time (ten minutes fast of Pittsburg), the Pittsburg operator in charge connects the main eastern wire to Philadelphia, 354 miles distant, with the observatory, and for the ensuing five minutes the beats of the Howard mean-time standard are automatically repeated on similar bells, or on the customary "sounders" in Philadelphia, and in every telegraph office through which the road wire passes--all station clocks and conductors' watches being compared with it as the official standard. After five minutes the clock is "switched" by the Pittsburg operator out of the main line wire, which is returned to its ordinary use.
    A similar set of signals, lasting for five minutes, is sent at 9 and 4 of Columbus time (thirteen minutes slow of Pittsburg time) to all stations as far west as Chicago, inclusive, in the main western line (of 468 miles in length). At Philadelphia the time is repeated to New York, at Harrisburg to Erie (333 miles), etc. As it is thus sent not only over the main lines from New York to Chicago (nearly a thousand miles), but over a number of subsidiary or branch roads too great for enumeration here, and which form in the aggregate a much larger number of miles than the main trunk, it will be observed that a considerable fraction of the railway system of the whole country is prepared for using a single unit of time; as, though the names of "Philadelphia time," "Altoona" or "Columbus time" are not yet abolished over that part of our railway system referred to every railroad clock and watch, and the movement of every train is regulated from a single standard--that of the clock in the observatory.
    The advantages of this uniform and wide distribution of exact time in facilitating the transportation of the country, and in enhancing the safety of life and of merchandise in transit between the Western and the Atlantic cities, seem to be sufficiently evident. The fact that the system described in this article has obtained the extension it has, within three years from its commencement, will, it may be hoped, justify the belief that its use has proved not only valuable to railways but an added security to the safety of the public.
    3rd. Supply of time to cities. At present arrangements are in progress for regulating the principal public clock of Pittsburg (the turret clock of the City Hall about two miles from the observatory), which it is intended shall strike every third hour on the bells of the fire alarm, and probably also in the various police stations. As the mechanism for doing this is still in course of construction, and may yet be modified in trial, it would be premature to speak of it, especially as its success has not yet been proven in practice here. The city clock will automatically report its own time to the observatory by a special wire, and it is probable that in controlling its rate from the observatory the "Jones" system will be used.
    The necessity of a uniform standard of time over the whole country, which was alluded to in the outset as one of growing importance, has not been further directly touched upon in this article, which is yet as a whole devoted to describing the means of meeting it. The evident tendency, in thus sending the time from one standard over so large an extent of territory, is to diminish the number of local times, and so prepare the way for a future system, in which, at least between the Atlantic and the Mississippi, they shall disappear altogether.
    A step in this direction has been contemplated by the managers of the roads uniting New York, Philadelphia, Pittsburg and Chicago, who have intended to use the time of the meridian of Pittsburg between the two extreme points mentioned, running all trains from New York to Chicago by this time alone, in place of using successively the local times of Philadelphia, Altoona and Columbus, as at present. Such a change would have already taken place during the last summer, except for an unexpected cause of delay, on whose removal it will be effected.
    The labors of this and of other American observatories are tending to the same important end--that of the ultimate adoption of some single time for the country east of the Mississippi, by which not only the railroads but cities and the public generally will regulate themselves. What point shall be chosen is of less importance than that some one should be used and universally.
    The subject is one which has hitherto attracted little public attention, but it does not seem unsafe to make the assertion that the causes which have almost insensibly effected such a revolution in England, will in a few years more bring it about here.

    Allegheny Observatory, Allegheny, Penn.,
                        Sept. 22, 1872,

American Journal of Science and Arts.