506 



SCIENCE. 



[N. S. Vol. III. No. 66. 



tin soon after the plates were taken, in 1872 

 and 1874, a simple comparison with the 

 new measures out to show whether the 

 plates still admit of accurate measurement, 

 and whether the positions of the star im- 

 ages have changed by an appreciable 

 amount. It is to be noted of course that 

 the Eutherfurd plates were made by means 

 of the wet-plate process, using albumenized 

 plates; so that the results of the present 

 paper are not strictly applicable to the 

 modern gelatine drj'-plates. Yet it seems 

 fair to suppose that the gelatine plates will 

 be at least as permanent as those of Ruth- 

 erfurd. In any case, the present research 

 is of considerable importance because of 

 the large number of Rutherfurd plates not 

 yet measured, and the measurement of 

 which would be useless if their precision 

 has been seriously impaired. 



It is therefore a source of congratulation 

 that the new measures here described have 

 not brought to light any such alterations of 

 the photographic film as would invalidate 

 measures made on the Rutherfurd plates 

 twenty years after the date of exposure. In 

 fact, we may say that in no instance does the 

 difference between the new and old meas- 

 ure exceed such an amount as might reason- 

 ably be expected from the combined uncer- 

 tainty of both. For the present purpose, I 

 have not thought it necessary to re-measure 

 all the plates treated in my paper on the 

 Pleiades (Annals N. Y. Acad, of Sciences, 

 Vol. 6, p. 239). Nor have all the stars 

 been re-measured, since a few stars well 

 distributed on the plate would undoubtedly 

 bring any existing change to light. On the 

 other hand, every care possible has been 

 taken to make the measures as accurate as 

 possible, except that the insignificant ' pro- 

 jection error ' found by Donner to exist in 

 the Repsold apparatus has not been taken 

 into account. Of course this is of no im- 

 portance in the work under consideration, 

 because the elimination of the errors of pro- 



jection would be almost certain to improve 

 the average accord with the old measures. 

 The same is true of any errors which may 

 perhaps exist in the guiding cylinder of the 

 Repsold machine, and which have also been 

 neglected. 



To avoid any possible bias in selecting 

 plates for remeasurement, I determined to 

 measure those plates to which even num- 

 bers had been attached by Rutherfurd at 

 the time the plates were made. But we 

 were unable to find plate number 20 among 

 the plates deposited at Columbia College, 

 so the remeasurement has been applied 

 only to plates 16, 18, 22 and 24. On each 

 of these plates eight stars were selected for 

 remeasurement, distributed on the plate in 

 a way well suited for bringing any disturb- 

 ance of the images to light. After this 

 work had been finished, it occurred to me 

 that the stars selected were all fairly bright, 

 and that it would be very desirable to 

 measure some faint stars too. Accordingly 

 six faint stars were selected, and v/ere very 

 carefully measured on plate 16. The stars 

 Anon. 34 and 18 m. were used as standards 

 on all the plates. 



Inasmuch as the Repsold machine fur- 

 nishes rectangular coordinates, whereas the 

 Rutherfurd measures were in distance and 

 position angle, it was necessary to compute 

 the distances and position angles from the 

 measured rectangular coordinates, before a 

 direct comparison could be made. The fol- 

 lowing table contains the results of such 

 comparison. In every case the ratio 

 adopted for the quantity : 



Rutherfurd scale value 



New scale value 



was such as would make the sum of the dis- 

 cordances in distance between the new and 

 the old measures zero. Similai-ly, a con- 

 stant was applied to the discordances in 

 position angle, so as to make the sum of 



