14 J. RHEINBERG ON THE ORIGIN OF CERTAIN COLOUR 



reflected and not transmitted light to render it visible.* Polari- 

 sation and refraction seem quite out of the question ; and as 

 there is no other theory at hand the answer must for the present 

 be left undetermined." 



Now, after hearing this paper my peace of mind was dis- 

 turbed. I am afraid I resemble in one respect certain animals 

 which go a little mad when a patch of colour is dangled before 

 them and they cannot make out what it is. At all events, I 

 thought I would see if I could not track out what this monstrouslv 

 irregular behaviour of A. Ralfsii was due to. 



The first thing to do was to ascertain how and in what way 

 the colour was formed. For this purpose it was found con- 

 venient to use a i-inch objective with a Davis shutter {i.e., a 

 small iris diaphragm) over it, by gradually closing which the 

 clear transparent valves would don their coats of many colours. 

 I had an idea that the difference in the colonrs of various valves 

 might be due to the number of perforations in the silex, or the 

 size of the perforations. That was soon shown to be incorrect, for 

 with a 15-power eyepiece it could readily be seen that the only 

 difference due to the closeness or size of the perforations was a 

 vslight alteration in the depth of the tint unaccompanied by any 

 perceptible difference of colour. But it furnished a clue in a 

 negative way, for it led to the observation that in the centre 

 of the diatom there were usually a number of isolated and some- 

 times irregularly shaped perforations which coloured up exactly 

 the same as in parts where the perforations were closely set ; 

 and finally, by very gradually and carefully closing the iris, it 

 could be noticed that the colour first made its appearance within 

 and around the edges of each perforation. Thus in a valve which 

 would ultimately be blue, each perforation became bluish with 

 purple-coloured edges. By further reducing the aperture of the 

 iris the separate perforations would vanish, for simultaneously 

 with a darkening of the colours they would join and run into 

 the adjacent perforations and give rise to the typical coloured 

 raylike appearance. 



* Theoretically this is not qiiite right, because the colours of thin plates 

 can be seen by transmitted light. Practically, however, it is perfectly 

 correct, because these colours would be too faint at approximately per- 

 pendicular incidence of the light ; moreover, diatoms are mostly too thick 

 for the colours of thin plates to come in question. 



