718 



LIGHT AND LIFE 



9 .8 y 6 



FracUon of Rhodopsin Present 



Fig. fi. Log threshold of a rod monothromat plotted against 

 fraction of rhodopsin present. 



III. Retinal Noisf. and Excitation by Light 



We have seen that in light adaptation the absolute threshold rises 

 greatly. It rises as an exponential function of the fraction .v of pig- 

 ment bleached, and according to Fig. 6 



Id 



— I e^o 



(4) 



where /„ is the fully dark-adapted threshold. 



There are two distinct ways of regarding the increase in I„. Ac- 

 cording to one, the noise of the system has increased so that the unit 

 of signal must be similarly increased in order to have the same signi- 

 ficance. This is formally equivalent to wasting signal, as when electric 

 current is degraded into heat by shunting a galvanometer, so that it 

 may operate at high signal strengths. A more efficient way of re- 

 ducing the sensitivity of a galvanometer is by stiffening the suspension, 

 so that no signal is lost, but sensitivity is exchanged for improved 

 temporal resolution. Does the retina operate like this? When alter 

 light adaptation, the absolute threshold rises lOO-fold, are 99 per cent 

 of the quanta wasted (as Wald suggests)? Or do they contriljute to 

 an increase in space-time resolution and to signal/noise reliability? 



In Limulus Hartline has obtained results, which he has never pub- 

 lished, relating the rise of threshold in light adaptation to the num- 

 ber of quanta used. The latter was deduced from the frequency of 

 obtaining a response with lights of various intensities, as in the ex- 



