which all observers have found very complex, will become much simplified, owing to the elimination of many lines hitherto attributed to these metals, but subsequently proving to be due to the presence of impurities in them. It may not be uninteresting to detail a few considerations which induce me to form such a conclusion. Instances have occurred in which a well-defined line appearing in the spectrum of a metal has proved coincident with the longest line in the spectrum of an element newly mapped.

Then, again, suppose iron to be present as an impurity in an element which is being mapped for the first time, the longest lines of iron are first looked for, but it may happen that one of these lines is represented in the spectrum of the new element as very much longer than the other lines in the spectrum of iron, which have hitherto been regarded of about the same length.

Lastly, there are many lines in the spectrum of one element reversed in the solar spectrum, which are absolutely coincident with lines of the same relative wave-length and definition in the spectra of other elements. It may be that greater dispersion may in all these instances prove that these lines, instead of being absolutely coincident, may slightly graze one another; but my experience up to the present time leads me to suspect that these lines are due to the presence of common impurities, either in the shape of unmapped elements or of elements hitherto unknown.