and Unmeasured Wave-lengtiu 



405 



observations up to about 20,000 of Angstrom's scale. Beyond 

 this point we have continued the curve both by computation 

 and by graphical extrapolation. We do not disguise from 

 ourselves the danger of all extrapolations, although ours rest, 

 it will be seen, on a wholly different basis from the ones 

 depending on formulae derived from the visible spectrum 

 alone, since our curve has been already followed by direct 



1-5700 





























































1-5600 





M 



























































1-5500 































































































1-5400 











fJjH 























1-5300 



















H 



































•3 -4 -5 -6 -7 '8 -9 10 1*1 1*2 1-3 1*4 1*5 1*6 1*7 1'8 1-9 



Showing ;j=/(A.) for Eock-salt. 



observations until it is almost coincident with a straight line. 

 Up to this point then (within the limits of error already 

 elsewhere given) there is no doubt, and unless there is some 

 utter change in the character of the curve, such as we have 

 no reason to anticipate, a tangent from the last part will not 

 differ very greatly from the immediate course of the curve 

 itself, and will at any rate meet the axis of abscissae sooner 

 than the curve can. If we assume, then, the prolongation of 

 the curve to agree with this tangent, we evidently assume a 

 minimum value for all the wave-lengths measured by it, and 

 that is what we have done. 



We are not prepared yet to speak of these wave-length 

 values as exactly determinate, and they are here given as 

 first approximations. They are indeed sufficiently startling 



