810 
broken registrationlines serve to identify the wavelength in several 
points of the spectrum. 
4. The fitness of the above mentioned method: to measure propor- 
tions of lightintensities by means of the photographic plate, will 
depend in the first place on the uniformity of the plate itself and 
on the faults in its structure. The first condition of its fitness is the 
possibility to reproduce a blackening, obtained by a definite intensity 
and time, by lighting another spot of the plate under quite equal 
circumstances. Each sort of plate therefore, before its use, has to be 
submitted to the following test: a series of equal spectra, taken 
with constant intensity of light, with the same time of exposure, is 
constructed one beneath the other on one photographic plate. The 
spectra then are photometered perpendicularly to their longitude in 
several different places. For each spectrum separately the blackening 
may not vary by passing in this direction, which is stated by the 
deviation of the photometer remaining constant. Further the elevation 
of these constant pieces for the different spectra must be the same. 
If the plate is all right, the registration with the photometer must 
give an image where the blackened resp. the clear pieces are lying 
on two lines parallel to the line of zero-points. In this way the best 
suitable photographie plate: the Panchromatic of WRATTEN and Wrain- 
WRIGHT, was chosen. (Panchromatic to be able to continue the measu- 
ring of absorption as far as possible in the red). But this plate too 
proved to be far from perfect. An accurate measuring gave the result 
that the blackening in the spectra on the border of the plate is always 
greater than in the spectra produced in the middle of the plate. 
A similar result can be obtained by measuring the blackening at the 
borders and in the centre of a photographic plate which is lighted 
uniformly over its whole surface. This systematic fault *) was eli- 
minated as well as possible by repeating each spectrum at least twice 
on each plate at different distances from the centre *). 
In the second place the regularity of the photographed spectrum 
depends highly on the kind of developer. The conditions to put on 
the developer is that it produces an equal blackening without spots 
1) The Kodac factory was not able to give a good explanation of this pheno- 
menon; they think of a drying out of the borders. 
4) To obtain in our case as many spectra as possible on one single photographic 
plate, the lengths of the comparison-spectra on the side of the small wavelengths 
was reduced. The substance examined namely did not allow any light to pass 
beyond the wavelength A= 4700. One half of the photographic plate was there- 
fore covered, the other used, and vice-versa. So doing it is possible to take 
on one plate of 9 & 12 c.m. two rows of 15 spectra each. 
