34 



Mr. F. W. Dyson. 



in the expedition from the Royal Observatory, Greenwich. The spec- 

 troscope used is a four-prism quartz spectroscope, kindly lent by 

 Captain Hills. The length of the spectrum from h (X 4102) to the 

 limit of the hydrogen series (X 3640) is 40 mm., so that the scale is 

 about 10 tenth-metres to the millimetre. 



The spectra were measured with one of the astrographic micro- 

 meters of the Royal Observatory (a micrometer originally designed for 

 measuring the photographs taken at the transit of Venus) by com- 

 parison with a glass scale divided to millimetres. The errors of the 

 5-mm. divisions have been accurately determined in the course of 

 investigations of the errors of the reseaux used in the photographic 

 chart of the heavens. The errors of the intermediate divisions were 

 determined by Mr. Davidson. The value of one revolution of the 

 screw of the micrometer is approximately § mm. 



The wave-lengths were deduced from the measures by an interpola- 

 tion formula, derived principally from the following lines, whose wave 

 lengths are taken from Rowland's tables : — 



Ca 



3968 



625 



Ti 



3761-464 



Ca , 



3933 



825 



Ti 



3759-447 



Ti 



3913 



609 



CrTi .. 



. 3757-824 



Ti 



3900 



681 



Ti 



3741-791 



Mg .... 



3838 



435 



TiFe .. 



. 3722-729 



Mg.... 



3832 



450 



Y 



3710-431 



Mg.... 



3829 



501 



Ti 



3685-339 



Y 



3788 



•839 



FeTi .. 



. 3659-901 



Y 



3774 



473 







These lines are the strongest lines in this part of the " flash " spec- 

 trum. In some of the photographs a number of the strongest iron 

 lines were also used as lines of reference. On the photographs taken a 

 few seconds before the eclipse became total the iron lines are unsuit- 

 able as lines of reference, as in some cases both a bright line and an 

 absorption line are seen, and in other cases the lines have a grey 

 appearance, and are not sharp and clear like the lines given above. 



The wave-length of h is only derived from one photograph, and is 

 not determined accurately. The value obtained agrees with the result 

 given by Mr. Wright,* in showing a correction of 0*1 of a tenth-metre 

 to the value given by Rowland. 



The intensities of the lines are given somewhat roughly. With the 

 exception of the cases noted where other lines apparently interfere, the 

 diminution of intensity is sensibly uniform. 



A comparison has been made with the wave-lengths given by 



n 2 



Balmer's law, using the formula X = 3646*140 — — ~ 3 the constant of 

 # ' Asfcroph. Journ.,' rol. 9. 



