The difference thus shown is so small that it might be ascribed to accident, did not other and subsequent comparisons in a measure corroborate it. Buys Ballot tabulated the result of 118 years' observations, summing the temperatures for each day of the moon's age, and his result also showed a maximum of temperature at about the day of full moon, strictly speaking, a few days after. But Mr. Park Harrison, who has classified several series of observations made in recent years, arrives at a conclusion almost the opposite of those of Toaldo and Buys Ballot, From readings of the thermometer at Greenwich, between 1841-1847 and 1856-1864, at Oxford between 18561864, and at Berlin between 1820-1855, he shows that the maximum occurs six or seven days after new moon and the minimum about four days after full. Mr. Harrison's curves laid down from his tabulation are remarkably consistent, and the difference between the maximum and minimum is very decidedly marked, for it amounts, from the average of all the series, to 2.5° Fahrenheit; that is to say, the whole heating effect of the moon upon Europe at first quarter is in excess of that at last quarter by two-and-a-half degrees.† Professor Loomis, using five years' observations made at Philadelphia for a similar investigation, arrived at a result strikingly coincident with this of Mr. Park Harrison's. How are we to explain the discordance between Buys Ballot's conclusion and these? It may be that it is due to some slight difference in the grouping of the observations; or it may be that the periods covered by Harrison's and Loomis's groups are not sufficiently long to give a reliable result. Buys Ballot divided his 118 years into twelveyear periods, and if some of these had been taken separately, it is apparent from an inspection of his tables that they would have given evidence differing in some degree from that afforded by the whole number of years.

By either conclusion, however, the fact of lunar influence on temperature is pretty well established. Mr. Harrison's, it is true, appears paradoxical, for he makes the new and practically cold moon warm us, and the full and presumably hot moon cool us. His explanation, however, is philosophical: it involves a consideration that was put forth by Buys Ballot twenty years ago, to account in some measure for the small amount of lunar heat evidenced by thermometric investigations in proportion to that which we may suppose the full moon to shed earthwards. It is an accepted conclusion that the heat is of two qualities— * Buys Ballot, "Changements périodiques de Température, dépendants de la Nature, du Soleil, et de la Lune." Utrecht, 1847, p. 79.

"Monthly Notices of Roy. Astron. Soc.," vol. xxviii. p. 39.
"Proceedings of American Association for the Advancement of Science,"

dark or invisible, and luminous or visible—the former consisting of rays which have been absorbed by the lunar surface, and then radiated from it; the latter of rays coming as part and parcel of the reflected sunlight; and it is generally assumed that the former quality preponderates. Of course we can know nothing of the capacity of the lunar surface matter for receiving and radiating dark heat, but upon the premise that bad reflectors are good radiators, we may base a conjecture that this capacity is rather great, for the general surface of the moon can hardly be called a good reflecting one. Some of the craters and spots, it is true, shine under high illumination with what appears like metallic lustre; but these form an insignificant part of the whole area of the visible hemisphere, and between them and the general surface there is a vast difference of brightness. When we observe the moon telescopically, and screen the eye with a smoke-tinted glass, these bright spots glow like heated phosphorus, while the general surface of the moon presents the dusky appearance of brown paper. Assume the former to have even a metallic polish, and the reflective power of the latter comes out very low. If the one has less than this assumed lustre-and we can hardly expect the craters to be mirrors-the other must be dull indeed; and if so dull to the eye, it has doubtless a strong appetite for heat. Sir John Herschel inferred the temperature of the lunar crust, after its 300 hour-long day, to be far hotter than the boiling point of water on the earth.* The German physicist Althous from calculation deduced that it becomes heated to several hundred degrees of our thermometer's scale; † and Lord Rosse, as one of the results of his observations to be presently described, considered that the absorbed solar heat raised the moon's surface material to a temperature of about 500° Fahrenheit.

The invisible rays of heat being wholly or in part intercepted by transparent media, it is obvious that those of this quality which come from the moon do not reach the earth's surface, but are absorbed in the higher regions of the atmosphere. In this case they must be effective in evaporating high clouds, dispersing such as are light, raising and thinning those that are dense. Herschel ascribed to this influence the tendency of the full moon to clear the sky, which proverbs assert, and sailors and peasants believe in; and although an astronomical observer, Mr. Ellis of Greenwich,§ has refuted the grounds for such a belief, so far as ordinary cloud registers supply means of examination, it has been confirmed by Professor Piazzi Smyth's observations upon the high clouds over the Peak of Teneriffe.

Herschel, "Outlines," sect. 432. "Proc. of Roy. Soc.," xvii. 436.

+ Poggendorf, " Annalen," xc. 544.
§ "Phil. Mag." 4th ser. xxxiv. 61.

In the account of the famous experiments upon the mountain top made in 1856, this observer states that "on Guajura there appeared to be a strong tendency to upper clouds during several days preceding full moon, but on that night every particle of them disappeared: the lower clouds, however, were constant through the whole lunation. This does seem to confirm Sir John's idea; and to show, too, that the moon's heat, though effective at great heights, is entirely expended before arriving at the lower strata of cloud, 2,300 feet above the sea. The elevation of the upper clouds, which were apparently so effectually acted on, we had no means of accurately judging of; but I should suspect that it could not have been less than 15,000 feet.”

Now, Buys Ballot pointed out, in 1847, that this absorption of heat by the higher airs is a cause of cooling to those atmospheric strata near the earth, since the evaporation of lofty clouds and the consequent clearing of the sky must permit a freer radiation of the earth's heat towards space, or at least into the higher regions of the atmosphere. And, as we have said, he considered this as a cause for the small amount of lunar warmth felt upon the earth. Mr. Park Harrison adopts this hypothesis to explain the high temperatures which he shows to occur at moon's first quarter, and the low ones which happen near the last quarter; and he considers that the time at which we observe the greatest effect in the former case is that when the part of the moon turned towards us has been least heated, and in the latter case that when the visible hemisphere has been longest subjected to insolation.

We might expect to see traces of the clearing effect of the warm moon in tabulations of cloud registers according to lunar periods; but we do not find them. Professor Schiaperelli has arranged thirty-eight years' cloud observations made at Vijevano in the order of days of lunation, and has laid down a series of curves which show the relations of clear and cloudy days to the moon's age.t But he shows the sky to be clearest when Mr. Harrison's theory requires that it should be cloudy, and cloudiest when, to fit the theory, it should be most clear. Meteorologists may one day reconcile these apparent anomalies. The evidence on either side is not, however, entitled to equal weight. Cloud observations have not the same accuracy as those of temperature: there is no instrument wherewith to make them. Much depends upon the judgment of the observer, and much also upon what is understood by "a cloudy night," or any other verbal record, and much again upon the interpretation of numbers denoting proportions of sky-area covered with cloud.

"Phil. Trans.," 1858, p. 503.

"Memoirs of the Lombards' Institute." 3rd ser. Vol. i. Fasc. iii.

Height and density of clouds are unheeded. Thermometer readings, on the other hand, are more definite, and less subject to erroneous classification; and we would rather accept their testimony of lunar influence on terrestrial temperatures, indirect though it be, than anything not susceptible of instrumental measurement. It would be a great point if we could measure the moon's dark heat directly, but this there seems little hope of doing, unless, as Professor Smyth suggests, we ascend to the level of the high clouds which he saw evaporated by the full moon-an altitude of about three miles.

The visible heat rays, and those which are not intercepted by the atmosphere, we can now, thanks to modern means of thermometry, measure with some pretensions to accuracy. Whether the moon's light possesses perceptible warmth has been a point of anxious and tentative inquiry among astronomers and physicists for a century and a half at least. The first observer who recorded a trial of it was, I believe, Geminiano Montanari, who, in 1685, thought he had indications of lunar heat in an ordinary dilatation thermometer. The second was Tschirnausen, the famous burning-glass maker. He published, in the year 1699, an account of a wonderful double lens, which concentrated the solar rays so far as to make them melt and fuse metals. It was formed of a four-foot burning-glass, with another lens of smaller size behind it. In telling of a number of its achievements, he mentions, little more than incidentally, that he turned it upon the full moon, but, though the image formed in the focus was of great brilliancy, there was no sensible heat.* He does not say what thermometer be used. Six years after, in 1705, La Hire the younger made an experiment with a burning mirror, belonging to the Paris Observatory, of 35 inches diameter, and an air and mercury thermometer of the construction then recently proposed by Amontons. The bulb of this instrument was two inches in diameter, and when it was placed at the focus of the mirror the moon's reflected image just covered it. The height of the mercury was noted, and the condensed moonlight was kept upon the bulb for a considerable time, but there was no alteration in the reading. This one trial satisfied La Hire that the moon's light was heatless.†

We find no record of attempts at lunar thermometry during the hundred years following the date of that of La Hire. But in 1820 attention was again turned to the subject by Professor Howard, of the Maryland University in the United States. Thermometers had by this time improved in construction: the instrument used by Howard was a modification of the differential one proposed by Leslie. The condenser was a mirror thirteen

"Histoire de l'Académie," 1699, p. 90.

† Ibid., 1705, p. 346.

inches in diameter, and one bulb of the thermometer, previously blackened to increase its heat-absorbing power, was mounted at the focus. The reflector was opposed to the light of the full moon, and, to quote the observer's words, "the liquid began immediately to sink, and in half a minute was depressed eight degrees, when it became stationary. On placing a screen between the mirror and the moon, it rose again to the first level, and was again depressed on removing this obstacle. I repeated this experiment several times to satisfy myself, and some of my friends who happened to be present, that there was no fallacy in the conclusion of its being a positive proof of the calorific power of the lunar rays, and at the same time affording an evidence of the great delicacy of the instrument."* The eight degrees here indicated were not those of any recognised thermometric scale, but mere arbitrary divisions upon the tube, about a millimètre apart. Within a year or two Pictet repeated Howard's experiment, using a similar thermoscope, but his index remained unmoved under lunar influence; if it altered at all, it gave an indication of cold. Prevost, reporting these results, pointed out that a mirror apparently reflects cold when exposed to the clear sky, because it intercepts the earth's warmth, and leaves the thermometer free to radiate its heat towards the sky. He thought the same effects would be produced whether the reflector be turned to the moon or to any other part of the heavens; and he suggested the desirability of experimenters making this test a part of their lunar thermoscopic observations. Further, he remarked that heat might be experienced occasionally, on fine summer nights for instance, for then the upper air is warmer than the lower.† Howard's strongly manifested warmth may have come from this source; possibly Montanari's also, and some detected by Frisius in 1781. Professor Volpicelli, in some historical notes on lunar thermometry, mentions certain researches by a Mr. Watt, which seemed to confirm Howard's experience. This was Mr. Mark Watt, a member of the Wernerian Society. His experiments, besides being rather unphilosophical, hardly bear upon the point: they refer rather to some supposed attractive and repulsive actions exercised by the moon's light upon little discs of metal mounted at the ends of a balanced bar.‡ There is mention of heat from the lunar rays, but it is ambiguous.

The experiments so far prosecuted, while they evidenced the anxiety of physicists to settle the point at issue, only proved the inefficacy of the means which, in expansion thermometers, they possessed for the purpose. But presently came a revolu"Silliman's Journal," vol. ii. p. 329. "Bibliothèque universelle," xix. 35.

"Edinburgh New Philosophical Journal," xix. 122.