32 
other crystals, which are thick enough to intercept all 
the primary rays. Yet the intensity ratios are, to all 
appearances, nearly correct before the allowance 1s 
made, and become quite wrong afterwards. The 
diamond behaves as if, like the other crystals, it were 
quite thick. ; 
I have therefore renewed a search for an effect 
which I have more than once failed to find, a special 
absorption of rays which are undergoing reflection. 
Since the earlier attempts the apparatus has gained 
in sensitiveness and accuracy, and I now find that the 
effect is easily visible. That is to say, when the pencil 
of rays strikes the diamond at the proper angle for 
reflection there is a diminution in the amount trans- 
mitted. 
In the experiment as arranged at present a pencil 
of X-rays from a rhodium bulb passes through a slit 
one-tenth of a millimetre wide, and falls upon the 
diamond, which is mounted on the revolving table of 
the spectrometer. The rays that pass through the 
diamond fall afterwards upon a crystal of rock salt 
so placed as to reflect a pencil into the ionisation 
chamber. When the diamond is turned, a minute of 
arc at a time, through the angle (about 9°) at which 
the diamond itself reflects the principal rhodium ray, 
the intensity of the ray reflected by the rock salt drops 
in the ratio 100 to 70. No doubt this ratio could be 
increased by more accurate arrangement. 
The principal rhodium ray is really a doublet, the 
two constituents of which are separated by an angle 
of four minutes under these arrangements. The 
doublet is resolved not only in the pencil reflected by 
the diamond, but also in the absorption band occurring 
in the reflection from the rock salt. 
The effect is no doubt analogous to the selective | 
absorption shown by crystals of chlorate of potash 
(R. W. Wood, Phil. Mag., July, 1906). 
W. H. Brace. 
The University, Leeds. 
Experiments Bearing upon the Origin of Spectra. 
Ir has been known for some years that a stream of 
luminous vapour can be distilled away from the mer- 
cury arc in vacuo, the vapour still remaining luminous 
when it has passed far beyond the limits of the electric 
field. It is known also that this luminosity is quenched 
when the stream passes near a negatively electrified 
metal surface. 
I have from time to time attempted to extend these 
results to other less volatile metals, and have now 
succeeded in a large number of cases. 
A preliminary account of some of the more signifi- 
cant observations will be given, without dwelling on 
experimental details. 
In the case of sodium under favourable conditions, a 
very curious behaviour is observed. Where the dis- 
tilled luminous vapour leaves the lamp, and where, of 
course, it is most brilliant, the light is yellow, and is 
dominated by the D lines. Further on, it becomes 
green, and the lines of the two subordinate series out- 
shine the D lines. Finally, further still, the D lines 
again predominate. It would seem that if we repre- 
sent the intensity of each series as dependent on time 
by a curve, the curve for the principal series will cut 
that for the subordinate series at two points. It is 
not, however, easy to find a law of decay which seems 
physically probable, and will satisfy this condition. 
Another interesting effect is seen when the luminous 
stream is made to pass through a negatively electrified 
wire net. As in the case of mercury, the glow is 
partially extinguished. But, if the glow is watched 
through a spectroscope while the negative potential is 
NOW 2305, VOL..934 
NATURE 
[Marcu 12, 1914 
applied to the gauze, it is seen that the lines of the 
subordinate series are far more affected than the 
D lines. 
We may regard the distilled glow as due either to 
persistent vibration of the luminous centres originally 
excited in the arc, or to some subsequent interaction 
occurring in the gas, such as molecular association, or 
the neutralisation of ions. Whichever view is taken 
| (and neither view is free from difficulty, as I shall 
/ show in a more complete publication) we must attri- 
bute the action of the electrified gauze to its power 
of attracting and neutralising positively charged ions. 
On either view the experiment cited shows that the 
systems which gave rise to the subordinate series are 
not the same as those which give rise to the principal 
series. 
In the case of potassium, the development of the 
subordinate series in the distilled glow is very strik- 
ing, and the existence of a series relation between 
the lines is visible at a glance, since the series are 
not confused by extraneous lines. The photography 
of this spectrum will be undertaken, and it is hoped 
will lead to an improvement in existing knowledge of 
the series and their convergence point. 
Lastly, I will refer to the behaviour of the glow 
from magnesium vapour, Initially, the colour is 
green, dominated by the triplet b, and the green band 
of the ‘‘magnesium hydride”’ spectrum, upon which 
as a background b lies. As the vapour moves on these 
die out, but the blue flame line at 44571 survives much 
longer. The vapour was passed through a wire gauze 
screen. On electrifying this to —4o volts, all the 
features of the spectrum which have been mentioned 
were seen to diminish in intensity, but the effect on 
the blue line and on the bands of magnesium hydride 
was much stronger than the effect on b. The extinc- 
tion of the band spectrum of magnesium hydride is 
specially significant. Re J. STRwGE 
Imperial College of Science, March 9. 
Unidirectional Currents within a Carbon Filament Lamp. 
Tue following experiments are good illustrations of 
| the thermionic current, or Edison effect, in a carbon 
filament lamp, and require only such apparatus as is 
usually found in a laboratory. 
The type of lamp used is that having two large loops 
in the filament, with the middle of the loop fixed by a 
short wire fused in glass at the top of the lamp. 
If the terminals are earthed and a charged body, either 
positive or negative, is brought near the lamp, then 
the two leaves diverge like two leaves of a simple 
electroscope. The loops may touch: the glass bulb, 
and, if so, they spring back discharged. 
But if the lamp is lighted and a pointed rod, con- 
nected to a Wimshurst, gives a powerful positive dis- 
charge, the loops are not displaced, even if the point 
is close to the bulb. On the other hand, with a nega- 
tive discharge, even a foot or two away, the two loops 
of the filament rapidly and repeatedly strike the glass 
and spring back. Apparently this action will go on 
for a long period, if the point of discharge is continued. 
The action may be explained from the fact that the 
lamp acts like a valve, and that the current can pass 
in one direction only, between the hot filament and 
the interior of the bulb. There can only be a ther- 
mionic current of electrons from the filament to the 
sides, and when there is an equilibrium distribution 
the carbon is at a relatively high positive potential 
compared with the inner wall. If this equilibrium is 
disturbed, it is adjusted by a thermionic current only, 
in one direction, or by movement of the loops only, in 
the other direction. 
Thus if the negatively charged plate of an electro- 
