554 HENRY A. ROWLAND 



Dr. Bell's paper s in which the calibration of the grating space is taken 

 into account. The method of correction is founded on the principle 

 that a linear error in the spaces only affects the focal length, and not 

 the angle, and that small portions which have an error, and thus throw 

 the light far to one side, should be rejected. The corrections Dr. Bell 

 has used seem to me very proper, except to grating III, which appears 

 to me to be twice too great. I find the following : 



Grating. D. Correction. Final Values. 



Mean value, 5896-20. 



This is very nearly the value given by Dr. Bell. 



The determination of Mr. C. S. Peirce of the TJ. S. Coast Survey is 

 certainly a very accurate one. Dr. Bell and myself have made some 

 attempts to calibrate his gratings, which he sent to us for the purpose, 

 and to correct for the scale used by him. There is great uncertainty 

 in this process, as we had only a portion of the necessary data. The 

 correction of his scale was also uncertain, because the glass scales used 

 by him may have changed since he used them, in the manner thermom- 

 eter bulbs are known to change. Correcting, then, only for the error of 

 ruling in the gratings, we have: 



Peirce's value 5896-27 



Correction * -07 



5896-20 



The correction for the scale would be about as much more in the same 

 direction, provided the glass scales had not changed. But it is too 

 uncertain to be used, although I have applied it in my preliminary 

 paper. 



Kurlbaum's result, made with two good modern gratings, has the 

 defect that the gratings were 42 and 43 mm. broad, quantities which 

 it is impossible to compare accurately with a metre. His small objec- 

 tives, one inch in diameter, could not take in light from the whole 

 grating, and so the grating space was not determined from the portion 



8 American Journal of Science, 1888. 



9 Bell, American Journal of Science, May, 1888, p. 365. 



