of the Spectral lines of the Elements. 197 
begin and end with a multiple of ten upon the scale. This has 
been done in Table III.* 
Taste IL 
H H’ b ¢ d 
589°5 818 656°5589 —19°0571 —0'4300 04984 
800 1018 619°548 —15°6660 - 03960 +0°4264 
0 1413 589-9750 —12°5440 +0-2369 +0-0165 
1413 1600 543-5600 — 78810 —0°8653 +0°2469 
1600 2200 527°3240 81670 OG OOTG oa ce 
2147 2315 489-5078 — 5°6093 “bap iGo) ea 
2230 3597 485°2950 — 6°1092 +0-2398 061 
3272 3773 098 — 3°3028 +0°1019 0°0043 
3597 4671 430-9556 a 9°7232 A 0OERS "  onenka 
TABLE IIL. 
H H’ b c d 
590 800 656-4 4637 —19°0614 —0 4225 10-4984 
800 1000 619°5488 — 15-6660 —0°3960 04264 
1000 1410 589-9750 —12°5440 +0:2369 -+0°0165 
1410 1600 543°795 — 78284 —0°8875 +0°2469 
2200 527°3240 81670 pPORTG 8 Eek. 
2200 2310 486°5804 — 54374 oe ee 
2310 3270 480°5580 — 5°7372 +0°2252 —0-0 
3270 3770 440°8759 — 3°3069 +0-1022 00043 
3770 4670 426 eons — 2°5734 ae has. 2 exe cc 
a be compared with those given in my paper on the a 
ment of Rete ths by the method of comparison in 
which Kirchhoff’s scale is subjected to a similar treatment, = 
Pe. required. In two cases a 8 line represents the ob- 
Pog best, the probable error is however unusually 
large. In the case of Mr. eases’ scale, on the other hand, 
pe robable error for each group appears to be much within 
mits of the errors of observation. In the 6th and 7th 
eee of data discussed in the eg paper, the sehen be- 
an | co 
not large, the curves here approac 
greater reoularity or relatively | lower order of the curves in the 
case of Mr. Hugeins’s scale arises doubtless from the fact that 
the prisms employed were, — whole series of observa- 
tions, in a constant position, which was not the case case with 
Kirehh hhoff’s apparatus. 4 
i Table ILI for computing wave-lengths the value 
a oe Wa subtract trecting the — il number in column 1 from the scale num- 
