NO. 21 COo ASSIMILATION AND WAVE LENGTH HOOVER II 



for the various filters represents the real distribution at the time of 

 observation. This factor may account for a large part of the vari- 

 ations in the original observations. 



In order to form an estimate of the importance of this source of 

 error for the corrected curve, I computed two sets of energy distri- 

 bution curves, one for a hazy day and the other for a clear day. 

 Following the method given above, two forms of the corrected curve 



3500 4500 5500 6500 7600 



WAVE LENGTH 



Fig. 4.— Wave-length assimilation curves. An, the corrected form of the curve 

 obtained with the large Christiansen filters ; B4, the corrected form of the curve 

 obtained with the small Christiansen filters. Points marked X, the results ob- 

 tained with the line filters and quartz mercury arc. 



were obtained. The maximum difference between these curves was 

 about 3 percent. This difference represents two extreme cases ; thus 

 we may assume that the corrected curve would not be changed by 

 more than half the error mentioned. This error added to the error 

 of the original measurements would indicate that curve A3 represents 

 the true form of the photosynthesis-wave-length curve to a probable 

 error of about 4 percent. The curve B4, as well as the values obtained 

 with the line filters, agree with curve A3. The difference is 5 percent 

 or less except in a short wave-length interval near 5800 A. Values 



