1G Dr. E. N. da C. Andrade on the Electrical 



at atmospheric pressure. Gr. Ebert * has observed that the 

 amount of deflexion of the streak in an electric field depends 

 markedly upon the amount of salt vaporized : he records 

 that with diminishing amounts of salt vaporized in a given 



time the deviation of the streak, after at first diminishing 



rapidly, finally becomes constant for amounts of salt less than 

 a given value. As this constant value was of particular 

 interest, Ebert confined himself to examining the effect with 

 small amounts of salt. 



It seemed to me of interest to study the phenomena with 

 very strong streaks, i. e., with a large amount of salt present, 

 and consequent rapid vaporization. Under these circum- 

 stances a new phenomenon appears : part of the streak is 

 deflected towards the positive electrode, and thus must be 

 negatively charged. The negatively charged part of the 

 coloured streak is always smaller than the positive part, but 

 there is no doubt as to its existence. It exists throughout 

 the flame, and not in any particular zone : to investigate 

 this one of the electrodes was replaced by a net, which 

 facilitates observation. 



The question arises, when does this phenomenon of a negative 

 deflexion begin to make its appearance ? It seems to take 

 place to some extent even for a faint streak, i. e., the side of 

 the streak towards the positive electrode is less deflected than 

 the other side of the streak towards the negative electrode. 

 The deflexion of the flame as a whole towards the negative 

 electrode tends to hide this ; however, for a taint streak. 

 both sides are certainly deflected as it' positive, though one 

 slightly less than the other. As the amount of salt present is 

 increased, the deflexion of the negative side (side towards 

 the positive electrode) does not increase much, although, as 

 already remarked, the deflexion of the positive side rapidly 

 increases : (he streak is broadened out. With large amounts 

 of -alt we get a plainly marked negative deflexion, but the 

 negative part is always considerably smaller than the 

 positive. The streak is forked above. 



Thus there is no doubt, that we can get negatively-charged 

 luminous metallic atoms in the flame, jusi a- we can in canal 

 rays t. This can be explained, remembering that the electrons 

 which are freed in the strongly positive side el' the streak 



• (i. Ebert, Heidelberg Dissertation, 1911. 



t The atoms here Bpoken of are those forming the coloured streak : it 

 has been shown elsewhere that some of the invisible negative carriers 

 which are driven oul of tin' streak, and detected on a Btnp outside the 

 dame, are metallic. ( I '. N.da ( '. Andrade, Phil. Mag, Juno L912, p 



