The authors are now engaged in pursuing this inquiry into the consonantal sounds.

II. "On the Reversal of the Lines of Metallic Vapours." No. V. By G. D. LIVEING, M.A., Professor of Chemistry, and J. DEWAR, M.A., F.R.S., Jacksonian Professor, University of Cambridge. Received February 20, 1879.

Since our last communication we have continued our experiments, using the electric arc as a source of heat, in lime and in carbon crucibles as described before. Success depends on the getting a good stream of vapour in the tubular part of the crucible. This is easily attained in the lime crucibles, which quickly reach a very high temperature, but are very soon destroyed; not so certainly in the carbon crucibles, which are good conductors of heat. The latter, however, last for a very long time.

In our experiments with tubes heated in a furnace we used a small spectroscope with a single prism, which gave a good definition and plenty of light; but in the experiments here described we have used a larger spectroscope by Browning, with two prisms of 60° and one of 45°, taking readings on a graduated circle instead of on a reflected scale.

Both in the lime and in the carbon crucibles we have found that the finely channelled spectrum, extending with great uniformity from end to end, always made its appearance so long as the poles were close together. A few groups of bright lines appear on it. We have not at present investigated this remarkable spectrum further. In several cases we have observed the absorption lines of the metals put into the crucibles on this channelled spectrum as a background, but generally when the vapours in the crucibles become considerable, the channellings give place to a spectrum of bright lines on a much less bright continuous background; we have used generally thirty cells in the galvanic battery, sometimes only twenty-five, once forty.

The calcium line with wave-length 4,226 almost always appears more or less expanded with a dark line in the middle, both in the lime crucibles and in carbon crucibles into which some lime has been introduced; the remaining bright lines of calcium are also frequently seen in the like condition, but sometimes the dark line appears in the middle of K (the more refrangible of Fraunhofer's lines H), when there is none in the middle of H. On throwing some aluminium filings into the crucible, the line 4,226 appears as a broad dark band, and both H and K as well as the two aluminium lines between them appear for a second as dark bands on a continuous background. Soon they appear as bright bands with dark middles; gradually the dark line disappears from H, and afterwards from K,

VOL. XXVIII.

2 D

while the aluminium lines remain with dark middles for a long time. When a mixture of lime and potassium carbonate (to produce a stronger current of vapour in the tube) was introduced into a carbon crucible the calcium (?) line with wave-length 4,095 was seen strongly reversed, and the group of three lines with wave-lengths 4,425, 4,434, and 4,454 were all reversed, the least refrangible being the most strongly reversed, and remaining so the longest, while the most refrangible was least strongly reversed and for the shortest time.

Besides these reversals, which were regularly observed, the following were noticed by us as occurring in lime crucibles but with less certainty, perhaps only at the highest temperatures. Dark bands appearing for a short time and dwindling into sharp dark lines with wave-lengths about 6,040 and 6,068 (perhaps due to the oxide); a dark line replacing the most refrangible of a well-marked group of several bright lines with wave-length 5,581 (or possibly the brighter line 5,588); and the lines with wave-lengths 6,121 and 6,161 reversed simultaneously for an instant and reappearing bright immediately; and the line with wave-length 5,188 reversed. When aluminium was put into the crucible only the two lines of that metal between H and K were seen reversed. The lines at the red end remained steadily bright. When some magnesium was put into a lime crucible, the b group expanded a little without appearing reversed, but when some aluminium was added, the least refrangible of the three lines appeared with a dark middle, and on adding more magnesium the second line put on the same appearance; and lastly, the most refrangible was reversed in like manner. The least refrangible of the three remained reversed for some time; and the order of reversibility of the group is the inverse of that of refrangibility. Of the other magnesium lines, that in, the yellowish-green (wave-length 5,527) was much expanded, the blue line (wave-length 4,703), and a line still more refrangible than the hitherto recorded lines, with wave-length 4,354, was still more expanded each time that magnesium was added. These last two lines expanded much more on their less refrangible than on their more refrangible sides, and were not seen reversed. The bright blue line (wave-length 4,481) seen when the spark is used, was not visible either bright or reversed; and this seems to be in agreement with Capron's photographs, which show this line very strong with the spark but not with the arc.

The following experiments were made in carbon crucibles:

When strontia was put in the lines with wave-lengths 4,607, 4,215 and 4,079 were all seen with dark lines in the middle, but no reversal of any strontium line less refrangible could be seen. After adding some aluminium and some potassium carbonate to increase the current of vapour, no reversal of any strontium red line could be detected, though momentary cloudy dark bands were seen in the red when

fresh strontia was thrown in. Two dark lines were seen in the extreme red, which proved to be the potassium lines reversed (wavelengths 7,670 and 7,700).

With a mixture of barium and potassium carbonates the line with wave-length 5,535 was strongly reversed, and that with wave-length 4,933 distinctly so. When barium chlorate was dropped into a crucible, the four lines with wave-lengths 4,553, 4,933, 5,535, and 5,518, were reversed, and as they remained so for some time, it is probable that the action of the oxygen of the chlorate had nothing to do with the result. The last-named line (5,518) was the least strongly reversed.

To observe particularly the effects of potassium a mixture of lime and potassium carbonate previously ignited was thrown in. The violet lines of potassium, wave-length 4,044, came out immediately as a broad black band, which soon resolved into two narrower dark bands having wave-lengths nearly 4,042 and 4,045. On turning to the red end the two extreme red lines were both seen reversed. No lines of potassium between the two extremes could be seen reversed, but the group of three yellow lines were all expanded though not nebulous, and other lines in the green were seen much expanded. These observations on potassium were more than once repeated with the same results.

Using sodium carbonate only the D lines were seen reversed though the other lines were expanded, and the pairs in the green had each become a very broad nebulous band, and D almost as broad a black band. When sodium chlorate was dropped into a crucible, the pair of lines with wave-lengths 5,681, 5,687, were both momentarily reversed, the latter much more strongly than the former.

When a very little charred rubidium tartrate was put in, the two violet lines were sharply reversed, appearing only as black lines on a continuous light background. Turning to the red end, the more refrangible of the two lines in the extreme red (wave-length 7,800) was seen to have a decided dark line in the middle, and it continued so for some time. The addition of more rubidium failed to cause any reversal of the extreme red line, or of any but the three lines already mentioned.

On putting some lithium carbonate into the crucible, the violet line of lithium appeared as a nebulous band, and on adding some aluminium this violet band became enormously expanded, but showed no reversal. The blue lithium line (wave-length 4,604) was well reversed, as was also the red line, while a fine dark line passed through the middle of the orange line. On adding now a mixture of aluminium filings, and carbonates of lithium and potassium, the red line became a broad black band, and the orange line was well reversed. The green line was exceedingly bright, but not nebulous or reversed, and the violet line still remained much expanded, but unreversed. With regard to the green lithium line, we may remark that we have no doubt whatever that it belongs to lithium, and that there must have been some

mistake in Thalén's observation, which ascribed it to cæsium. We have never detected this line with caesium, which, on the other hand, seems always to give the characteristic blue lines, both in the spark and in the flame, as well as to give the same lines reversed when its vapour is used as an absorbent.

When metallic indium was introduced into the crucible, both the lines with wave-lengths 4,101 and 4,509 were at once seen strongly reversed, and so continued for some time. No other absorption line of indium could be detected.

It is apparent that the expansion of lines, so often observed when fresh materials are introduced, must be ascribed to increase in the density of the vapours, not to any increase of temperature. Moreover, the length of tube which reaches a very high temperature in the experiments above described is very short in the lime crucibles, and still shorter in the carbon crucibles, so that the reversing layer is also short in many cases. We are, therefore, directing our attention to the means of heating up a longer length of the tubes, either by introducing oxyhydrogen jets, or additional electric arcs one above another; and also to the introduction of reducing gas (hydrogen or carbonic oxide) to counteract the oxidising action of the air which is drawn in through the lateral openings.

The curious behaviour of the lines of different spectra with regard to reversal has induced us to compare the bright lines of the chromosphere of the sun, as observed by Young, with those that are reversed in our crucibles. It is well known that some of the principal lines of metals giving comparatively simple spectra, such as lithium, aluminium, strontium, and potassium, are not represented amongst the dark lines of Fraunhofer, while other lines of those metals are seen and an examination of the bright chromospheric lines shows that special rays highly characteristic of bodies which appear from other rays to be present in the chromosphere are absent, or are less frequent in their occurrence than others.

In the following tables the relation between our observations on reversals and Young's on the chromospheric lines is shown.