branches of the dark flash would have been obliterated by the other rushes following. The effect of halation and solarization was also considered, but rejected. There was thus but one way to account for the phenomenon, namely, that the flash must have given out light of a wave length much shorter than the wave lengths of visible light and with a power sufficient to render the portion of the plate struck by it nonsensitive to ordinary light. Such a flash would appear black on a partially illuminated background or be invisible.

Dark flashes have been observed by the writer on several occasions and only when raining very hard. They appear to the eye the same as the accidental image produced after looking at a bright flash. Such an image may be retained in the eye for quite a while after, but can not easily be confounded with a real flash.

Two other pictures of dark discharges have been obtained with the moving camera, but are not so prominent; they likewise show the first rush as black, but without a bright core and with no side branches; one is of a horizontal and the other a vertical flash.

Below is a table of the measurements of this flash. The angle of the lens was 60°. Time of rotation of camera, 1 revolution in 10 seconds. Width of plate, 127 millimeters. The width of the whole flash, 48 millimeters. Time for the whole flash, 0.624 second. The average distance between the rushes, 1.2 millimeters. Average time for each rush, 0.0156 second. The plate used was a Standard" ordinary plate. The contrast of the picture has been increased by means of a double-contact print with lantern-slide plates. The developer was "Rodinal." Time for developing, five minutes.

66

In the summer of 1905 a new departure was undertaken by the writer at the suggestion of the Secretary of the Smithsonian Institution, the object being to obtain spectrum photographs of lightning.

Spectroscopic examinations of lightning have been made by many, but most of these observations have been visual, which at their best can only be rough approximations of the number of lines and their relative positions. As far as the writer knows, only one institution in the country-the Harvard College Observatory-had undertaken any work in photographing the spectrum of lightning.

A crude apparatus was constructed, consisting of a camera with a 35-millimeter prism fitted in front of the lens, no slit being used, as a lightning flash is a relatively narrow streak of light yielding a practically parallel beam. By means of this arrangement a few photographs have been obtained, two of which are reproduced in figures 6 and 7. A spectrum photograph of a spark from a static machine, for comparison, is shown in figure 8.

The spectrum shown in figure 6 is from a vertical flash, the picture of which was obtained June 18, 1905. It was about 13 miles distant and was taken at the end of a storm of local character. The spectrum of this flash resembles that from the static machine in most of its details.

Figure 7 shows the spectrum of one of those horizontal meandering flashes often seen at the conclusion of a storm of long duration. It was obtained September 1, 1905. It differs considerably from figures 6 and 8, several lines being absent. The first line of this spectrum is probably the same as the eighth line of the spark spectrum.

No definite opinion can at present be offered by the writer as to the meaning of these changes of the lines in the spectra of different flashes; more material must be obtained before a positive statement can be made.

Below is a table of measurements of the three spectra and their probable relations. Several measurements have been taken, and the average are here presented.

It may not be out of place here to give a few brief hints as to the best method for obtaining lightning photographs, for the benefit of the large number of amateur photographers scattered over the world. Thunder storms are nearly always cyclonic in their character, their diameter varying. When they extend over large areas it will usually be found that most of the vertical flashes are at the circumference of the circle through which the storm is passing. In the central part of a storm the flashes are usually horizontal, or passing between two

strata of clouds, therefore the best time for obtaining photographs is either at the beginning or the end of a storm. At the beginning the wind is usually very strong, hindering the work, so that it is genererally best to wait until the front of the storm has just passed overhead and then to expose the camera from a window or other suitable place facing a direction opposite to that of the wind. In this way the camera as well as the person will be protected from the wind and rain. This rule holds good in most cases, although there are many exceptions, especially when the storm is local in character and of small extent. A person must be on the lookout at all times and note the direction in which most of the downward strokes appear and direct the camera toward them.

It is useless to expose plates when there is nothing but sheet lightning, for it will only result in spoiling them. Oftentimes there is a mixture of sheet lightning and a few scattered ground strokes at uncertain points. When that be the case, the chances for obtaining pictures are slight, the sheet lightning soon fogging the plates, the time for fogging in such cases being from five to ten minutes, depending on the frequency of the reflections. The best rule to follow is to wait until a favorable opportunity presents itself, when the flashes are about half a mile distant, then act quickly, have plenty of plates ready to insert in place of the exposed ones, and trust to luck. The best way of holding the camera when swung by hand is to place it close to the body, tilting it somewhat upward, so as to get as much as possible of the sky in the picture, and swinging the body from side to side. The time and angle of the swing can be regulated with a little practice so as to be fairly accurate, say one second to the swing, although the writer has found by experience that when a sudden flash appears in front of the camera the evenness of the swing will be somewhat disturbed, particularly if the flash is close to the observer, unless the person be in possession of unusually strong nerves. When pictures of horizontal flashes are desired the camera must naturally have an up and down swing. These flashes are usually less intense and the pictures of them sometimes require a very long time to develop.

The developer preferred by the writer is "Rodinal," being convenient and clean, although any good developer will do, particularly such contrast developers as "Glycin" or "Hydrochinon."

Different kinds of plates have been tried, but the writer has no special preference for any one of the standard plates. The orthochromatic plates do not seem to give better results than the ordinary ones. Films give trouble in developing, because each exposure must be developed separately.

Figure 9 shows the barometer devised by the writer, which he has. found very useful in the study of the variations of the atmospheric

pressure preceding and during storms. The instrument recommends itself on account of the simplicity of its construction, its accuracy, and the ease with which the readings can be made, less than one twohundredth part of an inch being easily read off without the aid of a vernier.

The instrument shown in the illustration consists of a glass tube, one-fourth inch inside diameter, 38 inches long, bent to an angle of 100° at a distance of 8 inches from the closed end, filled with mercury, and mounted with the open end dipping into a reservior, which in this case is 1 inch in diameter. The open end can also be bent up to form a siphon barometer. The scale is placed along the upper slanting part of the tube. A pointer is fastened to the lower part of the instrument, the use of which is to indicate when it is in an exact vertical position, a mark on the wall indicating this position. The surface tension of the mercury is overcome by gently swinging the instrument and returning it to its vertical position. The calibration may be done by comparison with a standard instrument or by actual measurements. The bend of the tube may be made at a different angle, but should not be much less than 100° owing to the surface tension of the mercury. With this angle the instrument will magnify about six times, which has been found by the writer to be sufficient for ordinary use. The range of the instrument described is about 2 inches, and if a greater range is desired the slanting portion of the tube may be made longer or a different angle of the bend be substituted, although the proportions given have been found by experience to give the best satisfaction. A tube with a small internal diameter does not give good satisfaction owing to the surface tension being greater.