15 



the raeasureiuents were inadt; U[)Oii il;\ys lUiriiii;- wliicli tlie fiiutuatioii of tlie 

 barometer from the normal was very small, generally not more than one or two 

 millimetres at the time of making the reading. The exception was on Novem- 

 ber 26, when the heiglit was abont 10 mm. below the normal at the time of 

 making some measurements upon the E line. These measurements have been 

 corrected for this difference^ but this correction has not been introduced in any 

 other other instance as it bus not amomited to more than u small fraction of a 

 second, and its efiect will doubtless be eliminated from the general mean, owing 

 to the fact that these small fluctuations were about as often on one side of the 

 normal height ;\s on the other. 



THE VALUE OF THE GRATING SPACE 



By far the most difficult quantity to determine with accuracy is the value 

 of the grating space. When angular measurements can be made with the 

 precision which it is believed has been reached in the.«e determinations, the 

 com[)arison of the space of one grating with that of another can be made with 

 a degree of accuracy probably exceeding that attainable by ordinary methods. 

 For this purpose the gratings need not be brought together, the length of a wave 

 of light being a common measure easily apjilied to both. The value of the 

 space of one grating may be made to depend, therefore, upon that of another, the 

 absolute length of which has been ascertained with great care, provided both 

 have been employed in precise angular measurements upon one and the same 

 line. 



Instead of this a direct comparison of the grating space with a standard 

 unit of measure may lie undertaken. Where the first method is jiossible it is 

 vastly more convenient and, in general, more accurate than the last. Although 

 at lejist two Metres of well determined leu'^th are available in Jajian, no com- 

 paring ajiparatus suitable for sucii wurk coidtl be obtaitKid and it was, therefore, 

 not only desirable but nece.«.sary to rely, if possible, u|)on the first method for 

 the deterrainatiou of the grating space. During the progress of the observations 

 I liad hoped and expected to br; able to make; such a comparison between this 

 grating and another wlios»; space liad been determined to a great degree of 

 precision. Mr. 0. S. I'eirce of the U. S. Coast Survey, lias been engaged in a 

 series of observatiims of great interest and vahu^ for the comparison of a wave- 

 length of light with a standard metre. The ojieralion includes, of course, the 

 accurate measurement of the angle of deviation of a ccrtaiti line, prnduced by 

 a grating whose sjcu-e is delinitely known. 



Unfortunately I have not lieen able to prociu'e tlie results of this investiga- 

 tion, indeed I liclieve the final comparison of tint grating with the rni'tre has 

 not Ixjcn entirely c(implef<'d. I am llnndVire unable to give tia; value of flu; 

 grating «jMice with that degrctt of certainty whi'-h was at fuxt e\picli^d and 



