400 MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. 



We may here confine ourselves to describing tlie metliod used to give a standard interval of 

 time. A Ivey wliicli closes a contact was placed so that it was struclc by the falling screen, and 

 the contact closeil at a given point, say wlien the screen had fallen 44.1 mm., or three tenths 

 second. A contact which broke the same circuit was placed so that it would be broken when the 

 screen had fallen 78.4 mm., or four-tenths second. The circuit was consequently closed one-tenth 

 second. The distances were calculated theoretically and controlled by means of a recording 

 tuning fork. 



When the chrouoscope was being regulated and when the experiments were being made, the 

 apparatus was arranged as shown in the figure. The current which controlled the chrouoscope 

 was from twenty-four large gravity cells in pairs, which were placed in a separate room. After 

 the electro magnet had been rewound with coarser wire fewer cells sufficed. The wires were led 

 from a battery, B', to a comuuitator, 0, and thence the current passes through a rheostat, E, 

 and could, when required, be sent by a switch, *S', through a galvanometer, (/. From the 

 rheostat the current was led to a room in another part of the building in which the subject was 

 placed. In this room the current passed through one or two sensitive telegraphic keys. From 

 these keys the current was conducted to the chronoscope and thence to the contacts of the gravity 

 chronometer and back to the commutator and battery. 



When the chronoscope was being regulated the circuit was closed, excepting the upper circuit 

 of the gravity chronoscope. The electric chronoscope was set in motion and the screen allowed 

 to fall by breaking the separate current (from four gravity cells) B", which supplied the electro- 

 magnet. When the screen fell to the contact (G on fig. 3) it closed the circuit, and the hands of 

 the electric chronoscope were set in motion. After the screen had fallen 100 a farther it struck 

 .the lower contact (C on fig. 3) and broke the circuit, and the hands were stopped. The time 

 recorded by the hands would not usually be 100 (T, but a time perhaps 10 a or even 50 a longer or 

 shorter. The strength of the current (or the teu.siou of the spring in the chronoscope) was then 

 adjusted until the chronoscope gave the correct time. The current from the battery being very 

 constant, the variation from day to day was small, usually not more than 2 ff. but it might be 

 considerable if a change in temperature occurred. Under the conditions present a change in 

 time of 1 ff was caused by a change in resistance of about 5 ohms. The time was adjusted so that 

 the chronoscope had a constant error of less than 1 ff. The average variable error of the chrono- 

 scope as controlled by the falling screen was usually less than 1 a. Thus, in one trial (June 18, 

 1892) the average variable errors in seven series each containing ten measurements were 0.9(3, 0.8, 

 0.42, 0.4, 0.G4, 0.G4, and 0.50 ff. Occasionally, however (about once in ten trials), the time was about 

 7 G too short. We were unable to discover the origin of this error, but it was probably due to an 

 irregularity in the teeth of the wheel which caught up the hands of the chronoscope. This error 

 would make all times given in this paper about 1 crtoo short, and would increase the mean error of 

 the series. As we are, however, concerned with a difference in times, the result would not be 

 afl'ected except by increasing the probable error to the amount shown in the tables. 



The errors of the chronoscope are thus small. The variable error is practically eliminated in 

 a, series of one hundred experiments. The constant error of adjustment would not affect a difference 

 in time when the ])rocesses were measured alternately and on the same day. The constant error 

 in the rate at which the chronoscope runs would also be practically eliminated when a difference 

 is taken. As a matter of fact, this error is very small. In a series of seven determinations it was 

 one two-thousandth of the time. There is a theoretical error in the fact that the chrcuioscope is 

 regulated for 100 g, and in measuring longer times the current would magnetize the magnet more 

 and the times given would be too long. The magnetism would, however, be nearly complete within 

 100 G, and the times actually measured were always in the neighborhood of 100 g. We had projjosed 

 measuring this error, if appreciable, but in the meanwhile this has been attempted in Germany' 

 with entirely negative results. A very serious inconvenience in the chronoscope, as supplied by 

 the makers, is tmmd in the fact that the bar regulating the rate sometimes allows two vibrations 

 between each tooth of the escapement, and the times recorded are only half of the true times. 

 This error we have, however, nearly corrected, as described- above; in any case it does not cause 

 other than inconvenience, as the false rate is betrayed by the tone of the instrument. 



'Kiilpe uml Kirschniann, Philos. Stud., vii, 1892. 



