(4ENEEAL DISCUSSION OF SPECTROSCOPIC RESULTS. 
4(i5 
to estimate the intensities correctly, the weaker component Ijeing h’al)le to he 
considerably under-estimated. 
Another difficulty occurs when single Fraunhofer lines have a compound origin 
assigned, such as Fe-Ti, &c., the proportion of intensity of each element in the 
“make up” of the dark line being unknown. In such cases the relative proportions 
of intensity in the cori'espondmg hash line may be cpiite dillerent or even revei’sed, 
the i)redommating element being in general the one which ascends to the greatest 
elevation in the chromosphere, not necessarily the one which iiredoininates in the 
dark line. 
In these circumstances it is mij)ossible to make anything like a complete or hnal 
comparison of intensities. The liest that can Ije done is to select for each element 
isolated lines which are least open to the siisjjicion of lieing made up of more than 
one line in the hash spectrum, and also lines of supposed single origin as given in 
Howland’s tables. 
Unfortunately, there are only three elements which have a sufficient niimbei' of 
lines in their spectra to he treated satisfactorily in this way ; they are iron, titanium, 
and chiomium. In the following tables I give the results for these elements, 
selecting 2If) Fe lines of PtOWLANu’s intensity 3 and upwards, 124 Ti lines of 
intensity 1 and upwards, and 157 Or lines of intensity U and over. 
1 hese are reju'esented in the hash spectra by 93 Fe lines, 39 Ti lines, and 25 Cr 
lines respectively. Ilie selection of suitable lines was made entirely from Howland’s 
table, and without reference to the hash sjiectra, so as to avoid liias in the selection. 
ItOWLANDS intensities of the solar lines are given in the hrst column of each table, 
and the number of lines selected iietween XA. 3500 and 5000 in the second column, the 
tliiid and fourth columns give respectively the nuinbers and percentages of the lines 
which are found as bright lines in the hash spectrum, the hfth column giving the 
average intensity of these lines. 
A glance at the hrst and last column of each table will show the general relation 
between the hash intensity and the dai'k-hne intensities for the tlii'ee elements 
consideied. The numbers indicating intensities for tiie briglit and dark lines aie not, 
of couise, directly comparable, since they depend on methofis of judging intensity 
which may differ widely in the two cases. It is a mere coincidence in the case of 
iion that the numliers representing the stronger lines practically correspond in the 
hrst and last columns. 
From the columns of jiercentages the genei'al laile is obvious, that the stronger the 
daik line oi an element, the more probable is its occiiii'ence as a bright line in any 
given image of the hash spectrum. Thus we hnd that of the Ti lines none are 
present in the spectra under discussion corresponding to Howland’s intensity 1, and 
the percentage of dark lines exceeding intensity 1 which are present as bright lines 
iiicieases with each increase of dark-line intensity up to intensity 4 ; of the 12 dark 
lines exceeding intensity 4, all are present in the flash. Similarly with iron, all of 
VOL. CCI.-A. 
3 o 
