498 PEOI’ESSOR W. N. HAliTLF.Y AND ME. H. DAMAGE ON THE SPPXTEA 
The temperature of the flame of the “over-blow” is therefore much lower than 
during the first period, when highly heated carbon monoxide is undergoing combustion. 
This will, at least in jmrt, account for the feeble character of the line spectra of the 
alkalies, &c., at this stage. Another reason for this is discovered in the work of Gouy 
(‘ C. Ik,’ vol. 83, p}). 70-2 ; ‘ Phil. Mag.,’ 1877, No. 2, p. 156), who found that with 
salts of lithium, calcium, strontium, and barium the maximum intensity of tlieir 
flame spectra is reached “before tlie flaine ceases to he reducing (for a co})per wire), 
and is followed l3y a rapid diminution. With a large excess of air the .spectrum 
disappears.” The facts ap})ear unfavourable to the opinion which attributes the 
bands of their spectra to oxides. “ AVith sodium, on the contrary, the brightness of 
the flame augments rapidly as it becon\es less reducing ; tlie maximum is })roducedat 
the instant it ceases to be so. and is followed bv a reduction of briolitness much .slower 
^ */ O 
than with other metals.” 
N(3W without the decomposition jjroducts of the tar in the lining of the converter 
the flame cannot be more than feel)ly reducing ; the only reducing gas is the hydrogen 
resulting fi^jm the decomposition of the water vapour in tlie air. Gouv’s results, 
therefljre, also in part explain the absence of the line and bands of calcium, and 
account for the feeljle character of tlie other lines and for the changes in the spectrum 
desci'ibed below.The quantity of fume ex])elled is very great in the latter part 
of the “over-blow”; it is mainly composed of oxide of iron, and yet no bands nor 
lines of iron are present in the sjiectruni. 
The brilliancy of the flame is so great towards the end of the “ blow ” that we 
were led to sujipose that it might be due in jiart to an oxide of phosphorus. It 
is practically impossible to collect a sample of the fume five from the slag, and hence 
no direct evidence can be obtained on this point. F. E. Thompson states (‘ J. I. and 
S. Inst.,’ 1896, No. 1, p. 464) that “ phosphorus towards the end of ‘ hot bloAA's ’ 
does not pass readily into the slag, and scrap must be added,” that is, to cool the 
bath. This indicates that at the highest temperatures, in jiresence of a large excess 
of molten iron, phospliorus jientoxide is not formed as readily as at lower tempei'a- 
tures ; tlie less acid lower oxide of |)lios})horus would probably pass away with the 
nitrogen and be converted into phosphorus pentoxide by the external air. 
* III the Mow recorded on I’late 17 the red lines of the lithium and potassium with the yelloM" lines of 
sodium were seen with the direct-vision spectroscope as soon as the vessel was turned up. The liaiids of 
manganese appeared very early, Init they Avere not shaiqi during the first four minutes; they increased in 
brilliancy thronghont the first part of the blow, that is to say, until the flame dropped. The lithium and 
potassium lines coniinned very bright thronghont the whole of this period. At the dropping of the flame 
the liands all disappeared, as jiractically also did the lithium and potassium lines, the yelloAv sodium linos 
alone remaining visible in the rveak continuous spectrum. The flame at this time Avas A’ery short and 
feeble. It lengthened sloAvly as the Moav proceeded, until it became about one-fourth to one-third the 
length of the longest flame of the first period, and the lithium and potassium lines Iiecame stronger. 
5Vhcn the fume began to issue in ipiantity it had at first a rich purjile colour by transmitted light, Avhich 
colour faded gradually, then became like broAvn umber, Avliile the flame turned denser and denser until 
before long it had become very lirilliant indeed. The appearance during the “ after-bloAV ” (third period) 
Avas similar to that at the end of the second jieriod. 
