ON THE MEASUREMENTS OF WAVE-LENGTHS. 
85 
Various metallic lines measured by other observers compared with the new 
measurements (continued). 
Hartley 
and Adeney. 
Thalen. 
COBNU. 
Liveing 
and Dewab. 
Leoocq de 
Boisbaudban. 
o 
Angstbom 
and Thalen. 
Aluminium lines 
4477-2 
4478 
3961-0 
3960-7 
, . 
3962 
3943-4 
3943-3 
3943 
3092-2 
3091-6 
3081-5 
3081-7 
Indium lines . . 
4510-2 
4509 
4511 
4101-3 
4101 
4101 
Lead lines. . . 
4386-4 
4386 
4386 
4245-2 
4246 
4245 
4061-5 
4062 
4057-5 
4058 
4056 
Bismuth lines 
4301-4 
4302 
4303 
4259-1 
4259 
. # 
4259 
4121-2 
4119 
4118 
Carbon lines . . 
4266-3 
4266 
3919-5 
39i9-3 
3875-7 
3876-5 
2993-3 
2995-0 
2967-5 
2968-0 
2836-8 
2837-2 
2836-0 
2836-3 
2746-6 
2746-5 
2640-0 
2640-7 
2511-6 
2511-9 
2508-7 
2509-0 
2478-3 
2478-3 
2297-7 
2296-5 
Errors. 
There are four sources of error to which the measurements from our photographs 
are open. The total error is easily eliminated, but it is difficult to give precise values 
to the individual errors. They are :— 
First —The lens of the collimator was uncorrected. It was focussed for white 
light, and the adjustment was unaltered during the photographing of the spectra. 
On this account the rays striking the grating would not be parallel to one another, 
but be slightly convergent. Further, the rays corresponding to the different fiducial 
lines would be, after reflection, diverging from points not in the axis of rotation of the 
grating. To this source of error is mainly due the correction we have to make in our 
measurements. 
Second .— M. Cornu found during his researches on the “ Ultra-Violet Solar 
Spectrum” (Annales Scientifiques de l’Ecole Normals, iii., 1874, second series, p. 430), 
that in order to make his numbers obtained by direct vision and by photography 
