Polarization of the Light of the Sky. 507 



(where the positive and negative regions are separated at a certain 

 angle). 



It is to be regretted that no observations are given, by the 

 accomplished observer in question, of light emitted from the 

 beam at angles obtuse to the incident direction. It thus remains 

 uncertain, so far as these observations go, whether the pheno- 

 mena are continuous through the two right angles on one side 

 of the beam, or symmetrical in the two quadrants. The only 

 indication we can gather is, that the neutral emission seems to 

 have disappeared altogether in stages beyond that in which the 

 neutral angle is 90°. 



It is of course conceivable that the neutral angle might have 

 continued to increase beyond 90°. And if it is true that it does 

 not, there is a strong presumption that symmetry in the two 

 quadrants subsists at least up to the beginning of Stage III. 

 The question is easily seen to be the same as whether, under 

 normal circumstances, in the sky, the points of Arago, Babiuet, 

 and Brewster are equidistant from the antisolar point and the 

 sun respectively. As I have said, it is clear from the observa- 

 tions that this is so to a rough approximation ; and only further 

 observations, or especially further experiments, can decide in a 

 more accurate manner. 



We may embody the results of these experiments in the 

 remark that there exists a relation between the polarization ob- 

 served normally and the neutral angle, of the following nature: — 



Normal polarization. 



1 



diminishes 







Neutral angle. 

 Stage I. 



increases 

 90° 



\ Stage II. 



The first and last relations may be called the terminal con- 

 ditions. 



Before leaving TyndalFs experiments, we may notice the re- 

 markable effects he alludes to on p. 438. 



Observing the ring-system in common air (normally to beam), 

 puffs of smoke or of steam are introduced : the brilliancy in 

 both cases is greatly intensified ; but in the case of the steam 

 the ring-system is reversed. The actual characters of polari- 

 zation are not given ; but as the normal polarization in common 

 air (dust) is positive, it is to be assumed that the smoke was 

 also positive, and the steam negative. In a note a suggestion 

 of Sir John HerscheFs is mentioned as to the cause being a 

 change from polarization by reflection to polarization by refraction. 

 The earlier writers might have used this phrase simply to de- 

 scribe the phenomenon. I do not pretend to offer any criticism 

 as to the merit of the suggestion ; but I must point out that the 



