OP ARTS AND SCIENCES. 37 



numerous nell-marked sharply defined bands. The bands are most 

 numerous and brilliant in the green, and these give the prevailing 

 tone to the spectrum. But there is one very striking yellow band, 

 and there are also several bands in the blue and violet. Then in the 

 red there is an interesting double band, the two members of which are 

 the same distance apart as the two D lines. In addition, there may 

 be also a more or less diffused spectrum, which in some parts cannot 

 be distinguished from the similar diffused spectrum of hydrogen, and 

 it is worthy of remark in this connection, as indicating the purity of the 

 material used, and also that the diffused fluted spectrum above referred 

 to cannot come from the material of the tube, that no trace of the 

 sodium line was seen. No accomit was taken of the diffused spectrum, 

 as it only appeared when the battery was unusually strong. 



In speaking of the diffused spectrum of arsenic, we do not mean 

 the same kind of diffused spectrum as mentioned above in connection 

 with nitrogen. The diffused arsenic spectrum appears to be com- 

 posed of innumerable faint lines, wholly independent of the other 

 more brilliant characteristic arsenic bands ; and we only use the term 

 " diffused " for convenience, to express that the lines -are very faint 

 and too numerous to measure. 



The arsenic employed had been carefully purified by sublimation, 

 and preserved under distilled water. We used for measuring the 

 wave-lengths of the spectrum lines the spectroscope described b) 

 Professor J. P. Cooke [Am. Jour, of Science, Vol. XL., Nov., 1865]. 

 In this instrument, the train of prisms can be adjusted accurately 

 to the angle of minimum deviation, which was observed in each case. 

 We used five flint prisms of 45° angle each, and to reduce the angular 

 measurements to wave-lengths, we employed the method described by 

 W. M. Watt in his " Index of Spectra." 



We, in the first ijlace, measured with care the angles of minimum 

 deviation of the most prominent Frauenhoffer lines, and verified and 

 somewhat multiplied the data by measuring also the angles for char- 

 acteristic lines of the hydrogen, lithium, sodium, thalium, and stron- 

 tium spectra. These we combined with the wave-lengths of the same 

 lines given by Angstrom, by ordi nates and abscissas in the usual way, 

 and the curve drawn through the points so determined was so regular 

 and of such small curvature, that it was easy to interpolate with 

 minutes of arc to five tenth-metres of wave-length, as usually expressed. 

 The instrument is capable of reading to five seconds of arc, and with 

 the full bank of ten prisms it would give the wave lengths to tenth- 

 metres with perfect accuracy. With the comparatively feeble light of the 



