286 FROST— OBSERVATIONS OF THE ECLIPSE. 



pictures per turn, and after some preliminary experiments, I decided 

 that it was safe to have the crank operated at the rate of two turns 

 per second, giving sixteen pictures per second. Inasmuch as about 

 half of the time was used in moving the film between exposures, this 

 would represent about %o of a second for the exposure. On ac- 

 count of the clouds still lingering over the sun, it was not possible 

 to obtain satisfactory pictures at the beginning and end of totality 

 for bringing out the delicate details of the flash. But it is perfectly 

 evident from the pictures that an excellent record of the successive 

 stages of the development of the flash would have been obtained if 

 the sky had been clear. This slide shows a few of the pictures taken 

 several seconds after totahty : it will be possible to notice the tips 

 of bright gamma still reversed. The next slide shows a few of the 

 exposures, successively of the 15th, i8th and 23d seconds after 

 totality. Over 2,000 impressions of the spectrum were obtained. 



Another and important advantage of this method is that it re- 

 moves all the uncertainty of making the exposure at the correct 

 instant to secure the flash. This has been a matter of difficulty 

 and consequent nervousness on the part of the observers, even if 

 the signal should be given by a person observing the flash itself 

 visually. By beginning to operate the machine half a minute before 

 the expected time of totality, and running it for a few seconds after 

 totality has begun, there could be no doubt about catching the phe- 

 nomenon at all the stages and hence precisely at the best instant. 

 Thus the history of the reversal of each line should be shown, and 

 it would be very dilTerent for those of high level and those of low 

 level. This could be demonstrated to an audience or to a class by 

 the use of the film itself, and the film of course could be measured 

 as well as an ordinary photographic plate. 



It is very easy to connect a chronometer with the machine so as to 

 impress a dot on the film every second or half second, so that the 

 precise instant of each exposure can be known. At Green River, 

 Mr. Blakslee of our staff operated the crank and he was to receive 

 the signal from me as I watched the spectrum with a spectroscope 

 for this purpose. However, I decided during the partial phase that 

 it would be safer to begin the exposure one minute before the pre- 

 dicted time of totality. As a matter of fact, owing to the cloud, I 



