Besides the widening of the lines due to pressure, there is something else which must be mentioned. While experimenting with the spark taken between two magnesium wires focussed on the slit of the spectroscope by a lens, the lines due to the metal were found to be of unequal lengths. Now, as the lines are simply images of the slit, the lengths of the lines depend on the length of the slit illuminated, so that in this case it appeared that the slit was not illuminated to an equal extent by all the colours given out by magnesium vapour, but that the vapour existed in layers round the wires, the lower ones giving more colours, and so also more lines, than the upper ones further from the wire, as is represented in Fig. 186; this is only meant to give an idea of the thing, and is not, of course, exactly what is seen. S is the slit of the spectroscope, P the image of one of the magnesium poles; the other, being at some little distance away, does not throw its image on the slit, and therefore does not interfere. The circles shown are intended to represent the layers of vapour giving out the spectrum; on the right the lower layers give A, B, and C, the next A and B, and the upper ones only B. Now we may reason from this that the layers next the poles are denser than those further off, and give a more complicated spectrum than the others; and also, if the quantity of vapour of any metal is small, we may only get just these longest lines. Of late, experiments have been made in England on other metals—for instance, aluminium and zinc, and their compounds; and it is found that, when the vapour is diluted, as it were, one gets only the longest line or lines; and in the compounds, where the bands due to the compound compose the chief part of the spectrum, the longest line or lines of the metal only appear. Now what is the application of this? In the sun are found some of the dark lines of certain metals, but not all; for instance, there are two lines in the solar spectrum corresponding to zinc, but there are twenty-seven bright lines from the metal when volatilized by the electric spark. Why should not these also have their corresponding dark lines in the sun? The answer is, that the non-corresponding lines of the metal are the short ones, and only exist close to the metal where the vapour is dense; and in the sun the density is not sufficient to give these lines. Here, then, we have at once a means of measuring the _quantity_ of vapour of certain metals composing the sun. It was thought that aluminium was not in the sun, as only two lines of the metal out of fourteen corresponded to black lines in the solar spectrum. It is now known that these two are the longest lines, and that aluminium probably exists in the sun, and zinc, strontium, and barium must also be added. These probably exist in small quantities, insufficiently dense to give all the lines seen from a spark in the air. [Illustration: