248 SCIENCE PROGRESS 



extended atmospheres," The phrasing is evidently that of a 

 cautious reasoner; those who care to read the whole address 

 will find ample confirmation of the suggestion that it is no 

 idle speculation that is being put before us but a conclusion 

 towards which we are urged from more than one side. Thirdly, 

 there is a direct test for the existence of a fog which has 

 been applied to the depths of space with apparent success. 

 We all know that the sun looks red in a fog, because the red 

 rays of light can penetrate a fog better than the more refrangible 

 blue rays. For a similar reason our electric lights, being bluer 

 than gas, suffer obscuration more readily in a fog. If, then, 

 there be a fog in space, the more distant stars ought to appear 

 redder than the nearer. The test, however, is not so easily 

 applied to faint objects : for one thing we lose the sense of 

 colour when the light is very faint. But there is one character- 

 istic of red light that is familiar to all photographers : it takes 

 longer to photograph it, unless we use special plates. We have 

 then merely to ask the question, does the exposure required 

 for the more distant stars increase in an unexpected way ? The 

 answer is certainly in the affirmative, though we must be 

 careful that there is not another possible interpretation. From 

 the very beginning of the work on the Great Star Map it has 

 been a serious and fundamental difficulty that, when the 

 exposure was doubled, the gain of faint stars on the plate was 

 not so great as visual observations would lead us to expect. 

 The expectation was founded on laboratory experiments, which 

 show that, within proper limits, a light half as bright as another 

 will give the same photographic effect if the exposure is 

 doubled. " Within proper limits " — here is the need for care : 

 the law breaks down when the light is very faint indeed and 

 we must be careful not to mistake a breakdown from this cause 

 for a cosmical phenomenon. The " proper limits " are still under 

 investigation but they have already been subjected to careful 

 scrutiny : a considerable research by Dr. C. E. K. Mees 

 and Mr. S. E. Sheppard (to quote a single instance) indicates 

 that the limit is reached, for such plates as are used in the 

 Great Star Map, at about fifteen minutes of exposure.^ Now 

 well within this limit — for exposures of a few minutes only— 

 we find that the difficulty of photographing faint stars is out 



' See Investigations on the Theory of the Photographic Process (Longmans), 

 p. 214. 



