GOULD — REDUCTION OF D AGELET S OBSERVATIONS. 7 



gives tho translation of the second reading into degrees, minutes, and seconds ; and the last, or 

 10th, column contains remarks or memoranda made at the time, often records of the ther- 

 mometer and barometer, and occasionally remarks upon the estimated rate of the clock. This 

 column is wanting in the observations published by the Academy, and the few notes and 

 the meteorological observations are to be found in the first column. 



The printed records have proved to be seriously affected with errors. Many of these 

 are probably typographical, but the large number of other kinds show that no steps were 

 taken to insure correctness of the copy. The manuscripts deposited by d'Agelet must have 

 been the rough originals, which had apparently received neither revision since the moment of 

 observation nor any addition, unless the column of reduction for the second reading was 

 subsequently formed. I suspect, however, that these values were also written in at the time 

 of observation, on account both of the hasty way in which the translation must have been 

 made, and of the peculiar character of many of the errors. The printed pages were evidently 

 set up from these rough notes, without scrutiny or criticism. 



Consequently, instead of a reference to the originals as printed, and of a long catalogue 

 of Corrigenda, it has seemed preferable to prepare the observations in some such form as their 

 author would probably have given them, had his life been spared, correcting such errors as 

 have been rendered manifest or probable by the processes of reduction. The principle has 

 nevertheless been rigorously followed of making no change, however slight, in any of the 

 immediate results of observation, without recording it in a marginal note. ' The correction of 

 errors in computation requires, of course, no such record. 



$ 5. THREAD-INTERVALS. 



The distances of the three wires from their mean have been determined from 140 obser- 

 vations of transits over all three, taken in two groups of 70 each, tho one near the beginning 

 of the series of observations, and the other near the end. The resulting values for the 

 equatorial intervals are : 



from the first group, I = + 22s. 903, II = -f 0s. 137, 111= — 23s.041 



from the second group, 22s.889, Os.119, 23s.007 



their mean, which has been adopted, being 



I = + 22s.89G, II = + 0s.l28, III = — 23s.024 



which are not likely to be erroneous by the hundredth part of a second. 



From these was constructed the following table, which has been employed in the reduc- 

 tions : 



(?) 



