MEMOIR V.] INTERFERENCE OF RADIATIONS. 91 



the following August I made many attempts to obtain 

 Minilar specimens, but in no instance could the extra- 

 violet protecting action be traced, though the analogous 

 action of the red, orange, yellow, green, and blue was 

 perfectly given. Supposing, therefore, that the differ- 

 ence must be due either to impurities in the iodine or 

 to differences in the method of conducting the experi- 

 ment, I tried it again and again in every possible way. 

 To my surprise I soon found that the negative effect was 

 gradually disappearing ; and on Sept. 29 it could no 

 longer be traced, except at the highest part correspond- 

 ing to the yellow and green rays. In December it had 

 become still more imperfect, but on the 19th of the fol- 

 lowing March the red and orange rays had recovered 

 their original protective power. It seemed, therefore, 

 that in the early part of the year a protective action 

 had made its appearance in the red ray, and about July 

 extended over all the less refrangible regions, and as the 

 year went on it had retreated upwards. 



Are there, then, periodic changes in the nature of the 

 sun's light? The absorptive action of the earth's atmos- 

 phere is out of the question : if that were the cause, the 

 character of these spectrum impressions should vary 

 with the hour of the day. Or is it not more probable 

 that these singular phenomena rather depend on inci- 

 dental changes in the experiment, such as external tem- 

 perature, variations of moisture, the color of the sky, etc.? 



Under proper circumstances there is no difficulty in 

 exhibiting the power which the less refrangible rays 

 exert in arresting the action of the daylight : under such 

 circumstances a daguerreotype impression of the sun's 

 spectrum yields all three of the varieties of suiface be- 

 fore alluded to. The plate in the less refrangible and 

 extreme violet region is imaffected ; a narrow space of 

 white separates these unaffected portions from the in- 



