AucustT 24, 1899] 
oxygen a little further, we find that the six series only 
after all pick up the oxygen lines seen at a low temper- 
ature, and that if we employ a high temperature to 
observe the oxygen spectrum, that is to say, if we use 
an induction coil, a jar and an air break, we find a 
very considerable number of lines indeed which have 
no connection whatever with the series. And we are 
face to face with this very awkward fact, that in the 
case of oxygen there are more lines which we cannot 
get into a series than there are lines in the six series 
which we have attributed to that chemical substance. 
Here, therefore, we begin certainly to get into diffi- 
culties. The inquiry is not so straightforward, the 
conditions are not so constant, as we might have 
expected them to be. 
Here then we have instead of three series twice that 
number, and these only account for about half the lines. 
Now, let us look still a little further. The next point 
is that in the case of other substances we have no 
principal series, but only two subordinate ones. This 
happens in the case of magnesium, calcium and strontium. 
We have only two series in the case of magnesium, 
two in calcium, and two in strontium. In all those three 
NATURE 
695 
certain number of these lines has been picked up to 
form the series, but we get numerous lines which have 
been left over after all attempts to sort them into series 
have been made. 
I have now to bring before you another consideration. 
We are dealing in the case of calcium and magnesium 
with arc temperatures, but I showed you in my first 
lecture that in the case of calcium and magnesium the 
all-important lines in the hottest stars were lines seen at 
the temperature of the spark. I have added these lines 
to the diagram, and you will see that there is not the 
slightest trace of those lines having been picked up in 
the series. So that the further we go, the more we seem 
to get away from that beautiful simplicity with which we 
began. I take you now to another group of substances, 
namely, tin, lead, arsenic, antimony, bismuth and gold, 
and I might mention more. No series whatever have as 
yet rewarded the many attempts of those who have tried 
to get those metals and non-metals on all-fours with those 
previously investigated. It remained for Kayser and 
Runge to point out that it looked very much as if this 
complete absence of series was connected with the 
melting point of the substances with which they had 
been dealing. So long as the melting 
point was low, as in the case of sodium 
and lithium, the normal three series would 
show at low temperatures ; and, further, 
there were no lines over. But, when 
you get to these substances with high 
melting points, there is no series at all, 
and of course it is suggested that there- 
fore there must be intermediate stages ; 
and that really seems to be a very valid 
suggestion indeed, and one which in all 
probability will enable us to get over 
some of the difficulties. They point out 
that in the case of lithium, sodium, pot- 
assium, &c., all the lines are picked up, 
and that in the case of copper, silver and 
gold the series pick up only a very small 
proportion. There seems, therefore, to be 
a progression of complexity with the in- 
creasing melting point with regard to all 
the metallic substances which have so far 
been examined; of course this consider- 
ation does not touch the question of oxygen. 
Oxygen is a gas, hydrogen is a gas in con- 
Fic. 7.—Map showing series and residual lines in spectra of calcium and magnesium. 
we have a first and second subordinate series, but no 
principal series, I have studied the lines of calcium and 
magnesium, in the same way that the lines of oxygen 
were studied to see how many of the lines are picked up 
by the series. In the upper part of the diagram we have 
the lines seen in the arc spectrum of calcium, and in the 
two next horizons we have the lines picked up in the 
first and second subordinate series. The next horizon 
gives the residual lines—lines, that is, which have not been 
distributed into any of these series. You see that there 
is a large number outstanding just as in the case of 
oxygen, and it is very important indeed to note that the 
two lines H and K, which are more conspicuous in the 
spectrum of the sun than all the other lines of the 
spectrum, have not been caught by any of these re- 
searchers into the series of calcium. Therefore, with 
a reduced number of series, we seem to be getting 
still further from the simplicity we began with in the 
case of some of the permanent gases like hydrogen and 
helium. The same thing holds with regard to magnesium, | 
the spectrum of which at the temperature of the arc has 
not so many lines in it as the spectrum of calcium. A 
NO. 1556, VOL. 60] 
| the case of lithium. 
sequence, of course, of their very low melt- 
ing points, and you know that quite recently 
it has been found possible to liquefy both 
of them. So that there must be something 
different in their case, and it seems extremely encouraging 
to find, therefore, that the same variation, the same break- 
ing away from the law which I pointed out in the case of 
some of the metals, should really occur also in a gas, 
because it seems as if we shall be able to explain the 
phenomena in both cases by supposing that there is a 
condition of greater complexity, and that when we follow 
up this line of greater and greater complexity, whether 
in a gas such as oxygen, or ina solid such as gold, we 
do not get the simple series, because at the temper- 
atures we employ we are still far from the simple con- 
dition which we can get at in some gases and in some 
metals with low melting points. The table gives the 
relation between the melting point and the percentage of 
lines sorted into series. Thus, in the case of barium with 
a high melting point we get no lines at all represented in 
the series ; then we gradually get up to Ioo per cent. in 
But then again, as in the case of 
oxygen, when we come to mercury, which is also of low 
melting point, instead of getting 100 per cent. we only 
get about 25 percent. of the lines represented in the 
series. 
