THE GRID STRUCTURE IN ECHELON SPECTRUM LINES. 17 



echelon action. Otherwise the regions between hnes 1 and 2 would 

 have been filled in with a structure along axis aa' similar to that 

 along pp'. 



With an echelon alone we have obtained only the weaker component 

 of Li X6104 as a single narrow line. We plan to cool the tube with 

 liquid air, thus sharpening the stronger component so that it will 

 no longer suffer the secondary action, to which the small satellite is 

 probably due. 



(j) We have no record of having observed in either echelon any 

 ungridded line of width greater than Ag. Either there exists (1) a 

 very narrow line, (2) an irregular series of such, as, for instance, in the 

 yellow mercury lines, (3) a line of width Ag, (4) a series of such (the 

 grid more or less complete) or (5) a broad, structureless image cover- 

 ing between one and two orders. And it appears extremely probable 

 that the "reversal" of the main component of HgX5461, noted under 

 certain conditions by several observers and often noticed by us, may 

 be modified by the entrance of secondary action due to the excessive 

 breadth of this component. 



(k) The retardation producing the primary maxima of a narrow line 



is proportional to 7i — l, while that of the light undergoing secondary 



action is proportional to 3w — 1. Thus the difference in retardation 



in case of the two actions bears the ratio to the retardation of the 



2n . . 2w 



primary of ;, which is a function of n alone. The value of r 



n— 1 n — I 



varies from 5.50 for X6563 to 5.37 for X4341 in echelon No. 2; and from 



5.48 to 5.35 respectively in No. 1. Since echelons are generally made of 



substantially the same kind of glass, any two having equal separation 



of primary orders will have equal separation of secondary maxima, 



because this separation is the same fractional part of the separation of 



the orders; but the values of Ag in t. m., varying with the dispersion, 



will, of course, differ in different instruments. 



We cannot state just why Ao = 5Ag. The measurements given 

 above indicate that this is so within the limits of experimental error 

 for both the violet and red regions. 



It would be interesting to assemble an echelon under water, press 

 the plates together and allow the superfluous water to drain off. This 

 process might vastly reduce the secondary action. If successful 

 Canada balsam might be substituted for water thus producing a 

 more permanent instrument. We plan to try this experiment shortly. 



