182 
In some cases the phenomena were only incompletely visible, but 
this can be always explained. 
The downward motion of the band sometimes happened to be 
absent. This is the case if the air current is very slow. The tran- 
sition of one modification to the other has already taken place before 
the introduction of the fog into the condenser. 
The upward motion of the band will be imperceptible, if before 
the entire transition of the fog, the precipitation has been such that 
the effect becomes too small to be observable. 
21. We have tested also a hypothesis, communicated privately 
to us by a friend, and which would afford a possibility of explaining 
the observed phenomena, discarding the assumption of two sal- 
ammoniac modifications. 
The orientation of a crystal depends upon the surrounding medium 
and may change with it. 
Would it not be possible that in the case of positive double 
refraction the gas surrounding the particles is different from that 
present in the case of negative double refraction? For instance hy- 
drochloric acid or ammonia gas in the first case, in the second 
air with traces only of the mentioned gases. If then the dielectric 
constant of the environment is not much different from that of the 
particles, a new orientation might ensue, which would expiain the 
phenomena. 
Indeed all the preparations which we used allow of an initial 
excess of either NH, or HCl; in the experiment with sublimation 
(§ 19) an excess of one of the constituents might be due to the 
difference of the velocities of diffusion of the two gases. But in 
this last experiment air must be abundantly present. In order to 
look for a possible influence of the surrounding medium, the experi- 
ment of § 19 was arranged somewhat differently. A current of air 
was passed over a solution of NH,, the gases then were dried, and 
afterwards introduced into the tube, which contained the hot sal- 
ammoniae and lastly into the space with the condenser plates. 
The excess of NH, in the gas delivered from the apparatus was easily 
shown. The phenomena were the same as those described in $ 190). 
A similar experiment was tried with HCI in excess. The phenomena 
‘emained the same. It is preferable to use instead of air passing over 
a solution of HCl, a current of pure hydrochloric acid, obtained by 
dropping sulphuric acid into hydrochloric acid. 
22. We have also established the fact that NH, or HCI gas in the 
