586 Proceedings of the Royal Society 
make a list of all the primes. The only logarithms above 1900 
truly given are those of 2417, 2879, 2903, 6379, 8599, and 9137 ; 
and, with the exception of the logarithm of 9479 which is unit in 
excess, all those erring by less than 10 are in defect. A list of the 
corrections exceeding 9 is subjoined. 
Numb. Corrn. 
Numb. 
Corrn. 
Numb. 
Corrn. 
1303, - 10 
4201, + 
28 
6659, - 
2494 
1579, + 10 
4409, + 
55 
05 
00 
1 
45 
2003, + 13 
5233, + 
10 
6827, - 
25 
2011, + 12 
5273, + 
10 
6883, + 
30 
2203, + 55 
5813, - 
245 
7001, + 
53 
2207, + 30 
6011, + 
14 
7109, - 
295 
2633, + 13 
6037, + 
10 
8011, + 
10 
3307, + 55 
6269, + 
15 
8069, - 
494 
3863, + 25 
6521, + 
14 
8353, + 
12 
3923, + 10 
6581, + 
14 
8819, + 
31 
4007, + 19 
6619, + 
29 
9403, + 
15 
The only error higher than the sixteenth place is in the 
logarithm of 4603, which should be 93974 instead of 93924. 
From this it is obvious that the mechanical part of the work 
had been carefully performed, but that the computers had been 
unskilled in the management of the final figures, so as to prevent 
the accumulation of small errors. The fact that almost all the 
errors lie in one direction points to the influence of some definite 
but erroneous bye rule. 
Finally, on examining the list of corrections given in vol. iv. of 
the “ Annales de FObservatoire,” by help of which, according to 
“Nature,” Ylacq is to be made “much more accurate than any 
new table could possibly be,” I find between the narrow limits 
from 20000 to 30000 two omissions, at 24580 and 26699, and two 
mis-corrections, at 26188 and 29163, in all of which M. Lefort has 
been misled by errors of calculation made at the Bureau du Cadastre, 
as is clear from the subjoined logarithms set down true to the 15th 
place — 
24580 *39058 18785 50435 
26188 -41810 23322 49959 
26699 • 42649 49953 49034 
29163 -46483 21978 49968 
