Polarization of the Light of the Sky. 505 



The description of the change of the direction of emission for 

 maximum polarization in (7) /. c. is as follows : — " As the 

 actinic cloud grows coarser in texture, the direction of" (emission 

 for) " maximum polarization changes from the normal, enclosing 

 an angle more or less acute with the axis of the illuminating 

 beam." As this stands it would convey the impression that 

 this direction changed gradually from the normal. There are 

 no detailed experiments with vapours on this point ; and they 

 would be very difficult to execute, on account of the influence of 

 the oblique incidence on the glass tube on the polarization. But 

 here step in the experiments on smoke, recorded p. 437 (Radiant 

 Heat). No doubt there is an assumption in supposing that 

 these experiments are comparable with those on vapours; all 

 one can say is, that the smoke experiments appear to correspond 

 with those on the vapour so far as direct observation of the latter 

 is possible^ and also generally in other respects. And, accord- 

 ing to the smoke experiments, there is no doubt that what takes 

 place is, not a gradual change from the normal position for 

 maximum polarization, but a fading away of the positive polari- 

 zation normally emitted, and an origin and increase of a negative 

 polarization, the greatest value of which is in light emitted at a 

 very acute angle to the beam. Tyndall generally estimates this 

 angle of emission with the beam at 12°. But from his arrange- 

 ments it seems doubtful whether this is not simply the smallest 

 angle that could be conveniently observed. This point requires 

 more attention. The observations seem to indicate an increase 

 of negative polarization close up to coincidence with the beam 

 — a result which would appear impossible from the symmetry. 



In (8) /. c. there is a slight difficulty. " In passing from 

 section to section of the same cloud, the plane of polarization 

 often undergoes a rotation of 90°." It appears from the next 

 experiment (air, benzol, and hydrochloric acid) that the action 

 occasionally distributed itself unequally through the tube ; and 

 certain portions presented white clouds (i. e. an advanced stage 

 of action), while others remained blue (i. e. in a comparatively 

 early stage) . The plane of polarization of the blue segments was 

 parallel to the beam (Stage I.), while the plane of polarization 

 of the white segments was at right angles to the beam 

 (Stage III.). The causes of this distribution of the action 

 where advanced and retarded points lie side by side, do not 

 concern us here. 



The experiments on smoke at p. 436 (' Radiant Heat ') are for 

 our present purpose of the greatest importance. In describing 

 them I shall avoid the phrase " neutral point." It appears to 

 involve sub silentio the assumption that the different points of 

 the beam possess different properties. This docs not appear to 



