The contrast of the middle tones would thereby be increased but the high- 
lights and shadows would be pushed out to regions of very low gradient and 
under magnification it would be found that detail was seriously lacking in 
these areas. Use of the softer paper would give a much more uni form 
gradient throughout the tone scale and would improve the highlight and 
shadow detail at the expense of less separation in the middle tones. The 
gradient in the middle tones is still greater than unity, however, and the 
loss here is ouite negligible as compared with the increased transfer of 
information at the ends of the negative density range. This print uses a 
more restricted density scale and hence looks less contrasty to casual 
inspection. However, the inexact nature of photographic terminology is 
illustrated by the fact that we have actually increased contrast over an 
appreciable part of the tone-scale by using a "softer" paper. 
As advocates of the softer print, however, we are still not in an 
entirely satisfactory logical position, because, although we have reduced 
the variations of gradient over the tone-scale, we have not even approached 
uniformity, and an appreciable part of the negative range is still repro- 
duced at less than unity gradient. While the highlight and shadow grad- 
ients are higher than before, does the detail thereby brought out ccunter- 
balance the loss of large-scale contrast? How far can we push this use of 
softer papers? There can be no hard and fast or unique answer to these 
questions, and the argument cannot be pursued into further detail without 
reference to actual prints. However, most workers who have seriously 
studied this subject will agree that there is a marked gain of total 
information and no serious loss of any kind if we avoid densities less 
. than say O.l or greater than |.3. (These limits are arbitrary and vary 
with different papers.) 
Since we cannot achieve uniform gradient, is there an optimum distri- 
bution of gradient as between highlights and shadows? Although the 
superiority of the negative to the print might suggest a uniform gradient 
of unity as the desirable ideal, it seems likely that a higher value 
would be desirable because of the losses inevitable in any photographic 
operation, and due in this case to light scatter and diffusion in the 
system negative/air space/positive emulsion/baryta layer. 
Many years ago (1942) the writer suggested that a uniform gradient 
might not be the optimum, and that the desirable paper characteristic 
curve should be complementary to the combined transfer characteristics of 
camera flare, atmospheric haze and negative emulsion. This could only be 
achieved in a very approximate way, e.g. by having the paper gradient 
increase progressively with density; exact matching is obviously impracti- 
cable. The general validity of the idea seems to be borne out, however, 
by the observation, based on much printing of air negatives, that the 
upper useful limit of the characteristic curve is located at or near to 
the inflection point where the gradient begins to fall below the gamma. 
In general, use of densities above this point gives inferior reproduction 
of detail, and while no systematic investigation has been made, some 
interesting conclusions may be drawn. 
In the first place, it seems to follow that there is no upper limit 
to the density to which prints may be taken so long as the gradient is 
maintained constant or preferably increased. 
  
  
     
     
   
    
   
  
  
   
   
   
     
    
    
   
   
    
  
    
  
    
    
   
- ai oum C SCORSA 
N DO DO OO DD WS CTY U a Lt fh COO f: 
5 0030