The idea occurred in embryo while preparing for the Japan eclipse of 1887, and was roughly carried out on the spot; more carefully by means of pneumatic tubes for the eclipse of 1889 in West Africa; and most perfectly of all by electricity for the last eclipse of 1896, visible in the northern part of the Japanese Empirethe Hokkaido, as the island of Yezo is officially called.

Having had meteorological observations made at a series of possible eclipse stations through its track for three years, by the kindness of the Imperial Central Observatory, Esashi in Kitami province was finally selected a little fishing village on the shore of the Sea of Okhotsk, as the place least liable to be cloudy on the 9th of August; and thither was our apparatus transported.

Early in the autumn of 1895 Mr. D. Willis James, an honored trustee and liberal benefactor of Amherst College, and his son, Mr. Arthur Curtiss James, a graduate in the class of 1889, member of the New York Yacht Club, and an enthusi

astic deep-sea yachtsman, generously tendered their splendid schooner-yacht Coronet to convey our expedition to Japan. Also their wise liberality provided abundantly for all other expenses of the instrumental equipment and installation of the expedition.

The Coronet proved more than ample for her new service. Here she is, under sail-133 feet over all, 27 feet beam, and nearly 160 tons burden. Although she has no auxiliary, her magnificent sailing capacity easily maintains her at twelve knots in a favoring breeze; and so great was her steadiness on the high seas that the expedition lost but little time on the cruise; some of the instruments were even devised and built at sea.

On the 5th of December, 1895, the Coronet set sail from New York, with her cargo of the heavier apparatus made fast in the hold. The cruise of 117 days round Cape Horn brought her to San Francisco, where, on the 25th of April, 1896, the expedition joined her and started for Yokohama. Its members were Mrs.

James and Mrs. Todd, Captain James, P.A.-Engineer J. Pemberton, U. S. Navy, Arthur W. Francis, E. A. Thompson, Willard P. Gerrish, Vanderpoel Adriance, M.D., and Frank Thompson. All winter long the novel instruments had been in process of construction and trial in the laboratories and workshops of Amherst College.

A fortnight's cruise brought us to Honolulu, where a delightful sojourn of equal length was made. On the 22d of June the Coronet reached Yokohama. Every courtesy of an enlightened government was bestowed upon the expedition. Among them were free transportation northward over the railway to Aomori, at the further end of Nippon, and a special steamer of the Nippon Yusen Kaisha (Japanese Steamship Company) to convey the expedition and its material round Cape Soya, the "Cape Horn of Japan," to Esashi on the northeast coast. This was the steamship Suruga Maru, Captain Yagi commanding. Briefest allusion must suffice here; for these and numerous other facilities accorded us, together with the names of the officials offering them, find abundant mention in the publications of the expedition.

Esashi was reached on the 10th of July, a month before the eclipse. The old school-house, our headquarters, formed an ideal station. Here were in all perhaps a score of people-besides many of those before mentioned, a body-guard dispatched by the Governor-General of Hokkaido; Oshima, from the Imperial College of Agriculture, as interpreter; Shirasaka, the obliging village officer and his wife; Berthold, the efficient mate of the Coronet; and Japanese photographers, mechanicians, computer, carpenter, cook, and 'riksha-runner.

Within twenty-four hours our temporary home was well and comfortably established, regular work upon the installation begun, and some of the instruments were in their final position. A few days later Mr. K. Ogawa, of Tokyo, joined us as chief photographer, accompanied by two assistants; and it is to his skill that most of the pictures in this article are due. Much of the time preceding the eclipse the weather was very favorable, and, through the constant and faithful exertions of all members of the expedition, our preparations pro

gressed rapidly towards the eventful culmination. Two squares distant from our station was M. Deslandres, in charge of the French expedition, with an unusual outfit of spectroscopes; and a half-mile away was Professor Terao, who headed the party from the Government observatory at Tokyo.

At anchor in the roadstead off Esashi, in full view from our station, lay the sturdy cruiser Alger, Captain Boutet commanding, by whose delightful courtesy several members of the Amherst Expedition were enabled to return quickly to Yokohama after the eclipse was over, and permitted the rare and highly appreciated experience of a three days' cruise on board a French man-of-war.

To return to the automatic instruments and the successive steps in their evolution. My attention was first directed toward eclipse research in 1878, when I was sent to Texas by Admiral Rodgers, of the United States Naval Observatory, to conduct a search for intramercurian planets during the memorable eclipse of that year. The corona, as it appeared to me, was a complex phenomenon of endless detail, which no photograph of that eclipse or of any other has yet portrayed as the eye sees it. Excellent photographs had, however, been obtained; but the number of cameras available for a manual routine, and no less the number of plates exposable by hand, impressed me as exceedingly meager, for a supreme occasion when the money value of a single second is often hundreds of dollars.

A multitude of considerations all indicated the desirability of a new method of dealing comprehensively with the exposure and shifting of plates. Clearly an automatic system was the only solution, for the following reasons: The time lost in changing plates by hand, and in repointing the telescope when its adjustment is accidentally disturbed; the desirability of a greater range in exposures, so that the development of the negatives can be varied; the necessity of a mulititude of originals for distribution among astronomers and scientific societies; the lack of information on several important points where the testimony of past eclipses is conflicting-whether small instruments may not be equally effective with large ones, whether reflectors are superior to

of carrying its burden of a chain of 24 glass plates 8 x 10 inches; while another was loaded with 150 smaller plates, one of which was to be exposed every second of the total eclipse.

In all, twenty photographic instruments were worked into the automatic system; and the completest of preparations, but for an unhappy, though not unforeseen, condition of the Kitami skies, would have given us over 400 exposures, with several types of reflecting and refracting telescopes, photographic doublets, a pair of spectroscopes, photometers, and polariscopes; together with a sufficient test of both the old and the new photographic processes side by side.

All along I have spoken of the commutator as if it were a perfectly well-known instrument. But its photograph adjacent shows it to be quite otherwise, although in part it bears a strong resemblance to the familiar music - box. Every portion of it had to be devised anew, carefully constructed, and tentatively tested for its novel office; by it the electric circuits of all the instruments were unerringly controlled.

The basis of the commutator is an old chronograph with a ten-inch cylinder, and it was charged with the important duty of closing the multitude of circuits for all the instruments, each and every one at the proper instant and accurate to a small fraction of a second. At the righthand end of the barrel is a coarse feed-screw, and held rigidly in gear with it by a spring is a half-nut attached to the bent arm leading upward to the sliding board to which the contact-comb is secured. The number of con

this comb is forty-eight. The barrel revolves like that of an ordinary chronograph, once in sixty seconds. As totality was not to exceed three minutes, the contactsprings or teeth of the comb were placed at a distance apart equal to three threads of the feed-screw. Small brass contactpins were screwed into the cylinder wherever a contact was required. To facilitate placing them, the barrel was mounted in a lathe, and a delicate spiral traced over its entire length, using the same feed employed in cutting the feed-screw for the contact-comb. Exact correspondence was thus secured. At ten-second intervals fine longitudinal lines were drawn across the barrel, and the number of each tenth-second continuously from 0 to 8640 was engraved at the point of its intersection with the spiral. By this simple device it was possible to locate the position of each

THE ELECTRIC COMMUTATOR

minutes.