244 
FISCHER. 
made to assume the two fixed positions, A and B (Fig. 4). 
Then, by making the opening take the two positions indi¬ 
cated, it was not only possible to determine when the points 
and their reflected images were in proper focus, but the size 
and the direction of the error could be determined; for, if 
the conjugate focus did not lie in the plane of the cross-wires 
and the opening were moved, it is evident that the image 
would apparently be displaced with respect to the wires, and 
the direction and amount of the displacement would deter¬ 
mine the direction and size of the error of focusing. 
Shortly after the receipt at the Office of Weights and 
Measures of the copies of the International metre allotted to 
the United States, a direct comparison of one of them, No. 21, 
with the Committee metre was undertaken for the purpose 
of settling the question of the absolute length of the Com¬ 
mittee metre. The ends of the Committee metre in the 
vicinity of the axis of the bar are less perfect than the re¬ 
mainder of the surfaces, and hence points on the surfaces 
which varied from 2 to 3 mm. from the axis were selected 
for comparison. It was decided at the beginning to make 
use of platinum points instead of spider threads, as was done 
by Mr. Tittmann, for the reason that the air of the vault in 
which the observations were to be made was quite damp, and 
would thus make it difficult to maintain the spider lines taut. 
Four groups of observations were made, the mean temper¬ 
atures of which varied from 3°.75 to 22°.34 centigrade, and 
with probable errors which ranged from 0.23 to 0.34 microns. 
While all the observations were exceedingly accordant, the 
value found for the Committee metre at 0°.0 centigrade dif¬ 
fered by 3*526 * from the value derived from the 1889 obser¬ 
vations, the values being as follows: 
1889.. 0. M. = 1 metre — 0*538 at 0°.0 cent. 
1895. C. M. — 1 metre + 2*588 at 0°.0 cent. 
The movable screen of Cornu was not used in either of the 
comparisons referred to, and it is therefore extremely prob- 
* One V = one micron - one-millionth of a metre. 
