11 



nicivureiiients ;i ch;ui<;e must lie maile it' hiy;licr imwcrs arc in uso in the line 

 observations. 



^\gain, it was foiiiul that it was luiadvisable to make numerous repetitions 

 of an anE;le. exce|it when the temperature was imusually constant, for to make 

 the teniperattire correction it was ck-sinilile to know the temperature correspond- 

 ing to each setting rather than to depeiiil upon temperatm^e observations made at 

 the beginning and end of a series. Finally the measured angle, that through 

 which the grating wa.s turned, was much smaller than the corresponding angle 

 in the first methnd and in conseipience the results, which depend upnu the sines 

 of these angles, are alVected to a greater extent by small errors. A glance at 

 tlie formula shows that it is desirable to make a as small and b as large as 

 |>ossible. With a reflecting grating a cannot be diminished beyond a certain 

 limit: that is. with a spectrometer of ordinary construction. An instrument of 

 what may justly l>e called "extraordinary" construction has recently lieen made 

 by Alvan Clark & Sons for rrincetim College U. S. A. In this instrunu^nt the 

 collimator and telescope are one, as far as the use of the same tube and object 

 gU\ss goes. It is obvious that this ingenious plan gives the maximum efliciency 

 in certain operations, and if some optical difficulties which present themselves 

 can be entirely overcome this instrument will doubtless take rank as the most 

 accurate existing means of investigaliou in this direction. 



Concerning (lie first method, which was finally adopted, there are one or two 

 difficulties, whicli appear to Ik* of considerable moment at first, but which are 

 fbund to be man:igeable in practice. Willi a reflecting grating it is, in general, 

 ni'cessary to swing the telescope through at least two thirds of the whole 

 eircnmfercncc in makiriu; a sintrle pair of readings. While in this case the 

 mass to lie moved is much greater than in the i)revious method, yet the excellence 

 of the construction of the instrument is such that very little is lost in the way of 

 fmootlmess of niotiiin and ih licacy of selling and nnicli is y;ained in the lirniness 

 and staliiiily of the instrument by the considerable weight of its various parts. 

 üeieatfcl cxaininationR liave sliown tliat the various movements arc entirely 

 inde|ieiident of each other and tlic motion of the telescope never produces any 

 Hcncible disturbance of the other |)arts. I'v the use of this method, advantage 

 cannol Ik; taken of tlic principle of rr-petition as it can in ihat of oblifpie in idence. 

 lint it is believed tliat most of the lulvantages of repetion have been scciuvmI and at 

 t)n' Wime time the objection to it has lieeu avoided. The ease aiicl rapidity with 

 V hicii (Ik; grating could lie mmle normal to the collimator was Kueh that 

 measurements could lie readily made over dilVerent ]iart8 of the circle in yii cession 

 r.nd tins the errors of graihiation eliminated as far as possible. The temperature 

 correction, wliich is veiy importnnt, was secured by reatlin;^ tin; thermometor 

 immediatidy jvfler eacli setting upon a line. I'"inally, it )nay be urged that wo 

 can ni'ver be cct.iin that Ihc grating is absolutely normal to liic collimator. 

 'I'liis is (juite true, but it is also true llv\t the error in this adjustnn-nf may Ije 

 considerable without sensibly iitFecting tlie result. 



