296 E. M. NELSON ON METHODS OF ILLUMINATION. 



2. Not to form an opinion entirely by spectroscopic results. 



3. Not to imagine that one screen is sufficient. 



This is a difficult subject to deal with because words fail 

 to convey even a rough idea of colour. Colours must be seen^. 

 but the following descriptions of screens which prolonged experi 

 ments have proved to be the best may be of assistance. For 

 daylight, a piece of peacock-green, worked down so that it is 

 not too deep in tint, is combined with a very light-blue glass not 

 deeper in tint than a rather pale lilac petal. For lamplight^ 

 a thicker piece of peacock-green is combined with a blue glass^ 

 somewhat of the tint of the blue flower {Centaurea cyanus)^ 

 common in cornfields. If these glasses * have been correctly 

 chosen the microscopist after half a minute's work will not be 

 aware of the presence of a screen. The best method of comparing 

 the depths of the colours of screens is to place them upon a piece 

 of white paper. 



The ideal screen for visual microscopical work is one which,, 

 filtering out the too pronounced red, softens down, but does not 

 entirely cut out, the orange and yellow lights. Twenty-five 

 years ago any screen which did not pass certain spectroscopic 

 tests, by absolutely cutting out all lights longer than a definite 

 wave length, was rejected ; now we know better. For example,. 



* I introduced and brought this combination to the notice of the Club in 

 1894. The tints were, however, deeper than those I now use, but in 1895 I 

 stated that " if they are too dark they will obliterate fine detail, and it will 

 be better to pass a wider band even should it contain some objectionable 

 rays." So sixteen years ago I had recognised the principle of not being 

 guided altogether by spectroscopic results. 



The spectroscopic values of these two screens are as follows : 



Daylight without any screen . . . 430 down to 680 



,, with daylight screen . . . 430 ,, 630 



Lamplight with lamplight screen . , 450 620 



This information for visual purposes is valueless, although it looks so 

 scientific, because it does not indicate how much the orange and yellow 

 lights have been toned down. Although the red is only reduced by 10 

 with the lamplight screen, yet the effect of this screen is very much greater 

 than it appears, owing to the toning down of the orange and yellow. 

 The only method whereby screens could be standardised, or rather their 

 value expressed numerically, would be by Mr. J. Lovibond's tintometer. 



