THE ABSOTJP'J'ror^ AND KMISSION OK All!. 21 



doubling the image of the, spectrum, the disadvautivge of I'endei-iug the operation 

 of focussing difficult, inasmuch as the image of the slit partakes of the same 

 enlargement; so that, for example, the breadth of lines of about 185 /</< will be 

 somewhat more than double the width of the slit, while lines of about the shortest 

 wave-length hithei'to pliotographed will be somewliat less than double the width 

 of the slit. The edges of such a broadened specti'al line cori'espoud accordingly to 

 two different focal distances. Tliis difference becomes proportionately more notice- 

 able the wider the slit and the shorter the focus. Since in my hydrogen photo- 

 graphs foci down to 70 mm. occur aiul a width of slit of 0.010 mm. coi-responds 

 nearly to a difference of focus of 0.02 mm., it might be concluded tliat under 

 circumstances so luifavorable no specti'a good for anything could be produced. 

 Nevertheless photographs of this sort are clearer than might be expected. The 

 reason is that with this relatively wide slit the scattering of tlie I'ays by diffraction 

 is of less importance than with nairower slits, although with the latter the disad- 

 vantage of the diffei-euce of foci of the two edges is not noticeable. I shall return 

 to the subject below. But a more serious difficulty arises from this cause ; and an 

 acquaintance with it is desirable in adjusting a vacuum-spectrograph for the diffei'- 

 ent photographic fields within its reach. The state of things is this: Before a 

 vacuum-spectrograph is set to its regular work, it should be calibi-ated. That is to 

 say, for each photographic field the following constants shoidd be definitively 

 determined : the focal distances, the minimum deviation, the angle between the two 

 optical axes, the angle between the sensitive plate and the optical axis of the tele- 

 scope. I distinguish, foi- my apparatus, nine such photographic fields beyond 

 185 /<y^. Suppose, now, that the obsei'ver has gone through with this arduous task 

 of calibrating his spectrograph for its entire range, and has performed it with 

 a somewhat wide slit ; for to save time in the exposures he will naturally prefer to 

 do it so. He will take, let us say, a width of 0.01 nun., and although his slit opens 

 unsymmetrically, he will assume that this same adjustment is applicable for 

 narrower slits. But with these he will find that the greatest sharpness of image 

 has not the focus found for the wider slit, but a somewhat different one ; that, 

 moreover, for the attainment of the highest sharpness of images the lenses can now 

 be a.ljusted incomparably more finely; and finally, that the useful part of the 

 spectrum has materially suffered in length. The principal cause of this phenomenon 

 is the oblique positiou'of the photographic plate with reference to the optical axis; 

 but the uusymmetrical variation of the slit-width, that is, the assumed fact that the 

 slit-carriage has only one movable ja^v, aggravates it. It might seem that the 

 difficulty could be obviated by a symmetrical slit. But little could be gained m 

 that way; since the symmetrical motion of the slit-jaws with a slit of a micron's 

 width will present to the instrument-maker insuperable difficulties, as I have 

 sufficiently convinced myself )>y extensive experiments with slits of tlie highest 



precision I could construct. • , ■ i i- 



The reason why, with a plate standing obliquely to the optical axis, the focus 

 dei^ends upon the wi.lth of the slit, appears from the following considerations: 

 Suppose the focus for any spectral line has been ascertained xvith a very narrow 



