1908-9.] Negative Attempt to detect Fluorescence Absorption. 409 
placing a cell with copper nitrate, of a suitable strength, as colour filter 
before the comparison source. In the measurements made on fluorescein 
and the other piece of uranium glass, which showed a stronger fluorescence, 
a narrower slit was worked with and this precaution was not necessary. 
In the article above referred to Wood has indicated a method of testing 
the method employed for systematic error. The test would be improved 
if a piece of coloured glass, instead of a piece of clear glass, were placed 
before the slit at an angle of 45° to the axis of the collimator. 
Fluorescein. 
A standard solution was prepared by dissolving as much fluorescein as 
possible in water at room temperature, the undissolved particles being 
removed by filtration. The following table gives the fraction of the 
incident light transmitted through 1 cm. of this solution : — 
A 
A 
535 fx/x 
•918 
484 ixjx 
•0117 
515 
•710 
475 
•0381 
504 
•168 
466 
•0725 
496 
•0305 
452 
•1441 
489 
•0070 
438 
•3514 
The above table does not include the loss for reflection. Thus for 
A = 515 /a/* '290 of the light entering the solution would be absorbed in it. 
The next table gives the fluorescence curves not corrected for absorption 
in the solution. In (1) the exciting source had not, in (2) it had the 
cobalt glass filter. A solution 1 cm. thick and one quarter the strength 
of the standard solution was used. 
A 
(i) 
(2) 
604 fifi 
'0187 /x/x 
• . • 
574 
•137 
•038 
549 
•476 
•175 
527 
1-37 
•516 
508 
2-30 
•881 
492 
1-08 
•486 
The curves are plotted in diagram 5. The results of two sets of 
readings made on A = 508 ju/u are given : — 
T 
F 
C 
T + F-C 
Probable 
Error. 
906 
290 
1173 
+ 23 
10 
935 
694 
1593 
+ 3 
20 
