I would next ask your attention to the fact that in either spectrum, below pi = 12,000 are most extraordinary depressions and interruptions of the energy, to which, as will be seen, the visible spectrum offers no parallel. As to the agent producing these great gaps, which so strikingly interrupt the continuity of the curve, and, as you see, in one place, cut it completely into two, I have as yet obtained no conclusive evidence. Knowing the great absorption of water vapor in this lowest region, as we already do, from the observations of Tyndall, it would, a priori, seem not unreasonable to look to it as the cause. On the other hand, when I have continued observations from noon to sunset, making successive measures of each ordinate, as the sinking sun sent its rays through greater depths of absorbing atmosphere, I have not found these gaps increasing as much as they apparently should, if due to a terrestrial cause, and so far as this evidence goes, they might be rather thought to be solar. But my own means of investigation are not so well adapted to decide this important point as those of photography, to which we may yet be indebted for our final conclusion.

[Illustration: FIG. 2.--NORMAL SPECTRUM. (At sea level.)]

I am led, from a study of Capt. Abney's photographs of the region between pi = 8,000 and pi = 12,000, to think that these gaps are produced by the aggregation of finer lines, which can best be discriminated by the camera, an instrument which, where it can be used at all, is far more sensitive than the bolometer; while the latter, I think, has on the other hand some advantage in affording direct and trustworthy measures of the amount of energy inhering in each ray.

One reason why the extent of this great region has been so singularly underestimated, is the deceptively small space into which it appears to be compressed by the distortion of the prism. To discriminate between these crowded rays, I have been driven to the invention of a special instrument. The bolometer, which I have here, is an instrument depending upon principles which I need not explain at length, since all present may be presumed to be familiar with the success which has before attended their application in another field in the hands of the President of this Association.

I may remark, however, that this special construction has involved very considerable difficulties and long labor. For the instrument here shown, platinum has been rolled by Messrs. Tiffany, of New York, into sheets, which, as determined by the kindness of Professor Rood, reach the surprising tenuity of less than one twenty-five-thousandth of an English inch (I have also iron rolled to one fifteen-thousandth inch), and from this platinum a strip is cut one one-hundred-and-twenty-fifth of an inch wide. This minute strip, forming one arm of a Wheatstone's bridge, and thus perfectly shielded from air currents, is accurately centered by means of a compound microscope in this truly turned cylinder, and the cylinder itself is exactly directed by the arms of this Y.

The attached galvanometer responds readily to changes of temperature, of much less than one-ten-thousandth degree F. Since it is one and the same solar energy whose manifestations we call "light" or "heat," according to the medium which interprets them, what is "light" to the eye is "heat" to the bolometer, and what is seen as a dark line by the eye is felt as a cold line by the sentient instrument. Accordingly, if lines analogous to the dark "Fraunhofer lines" exist in this invisible region, they will appear (if I may so speak) to the bolometer as cold bands, and this hair-like strip of platina is moved along in the invisible part of the spectrum till the galvanometer indicates the all but infinitesimal change of temperature caused by its contact with such a "cold band." The whole work, it will be seen, is necessarily very slow; it is in fact a long groping in the dark, and it demands extreme patience. A portion of its results are now before you.

The most tedious part of the whole process has been the determination of the wave-lengths. It will be remembered that we have (except through the work of Capt. Abney already cited, and perhaps of M. Mouton) no direct knowledge of the wave-lengths in the infra-red prismatic spectrum, but have hitherto inferred them from formulas like the well-known one of Cauchy's, all which known to me appear to be here found erroneous by the test of direct experiment, at least in the case of the prism actually employed.

I have been greatly aided in this part of the work by the remarkable concave gratings lately constructed by Prof. Rowland, of Baltimore, one of which I have the pleasure of showing you. [Instrument exhibited.]

The spectra formed by this fall upon a screen in which is a fine slit, only permitting nearly homogeneous rays to pass, and these, which may contain the rays of as many as four overlapping spectra, are next passed through a rock-salt or glass prism placed with its refracting edge parallel to the grating lines. This sorts out the different narrow spectral images, without danger of overlapping, and after their passage through the prism we find them again, and fix their position by means of the bolometer, which for this purpose is attached to a special kind of spectrometer, where its platinum thread replaces the reticule of the ordinary telescope. This is very difficult work, especially in the lowermost spectrum, where I have spent over two weeks of consecutive labor in fixing a single wave-length.

The final result is, I think, worth, the trouble, however, for, as you see here, we are now able to fix with approximate precision and by direct experiment, the wave-length of every prismatic spectral ray. The terminal ray of the solar spectrum, whose presence has been certainly felt by the bolometer, has a wave-length of about 28,000 (or is nearly two octaves below the "great A" of Fraunhofer).