which reaches it in the opposite part of its orbit, in the proportion of 932 to 902, or about as 1000 to 936. Midsummer of the southern hemisphere is the season when the earth is nearest to the sun; the winter of the southern and the summer of the northern hemisphere' occur when the earth is farthest from the source of heat. The conclusion seems inevitable-the southern hemisphere must have hotter summers and colder winters than our hemisphere, where the heat of summer is tempered by the greater distance, and the cold of winter mitigated by the comparative nearness, of the sun.

The next point to be considered is the effect of ocean-currents, and especially of the Gulf-stream, in modifying the climatal -conditions of some parts of the earth. Following in the track of the late Captain Maury and Principal Forbes, Mr. Croll has especially insisted on the importance of the great current which, issuing from the Gulf of Mexico, and flowing northward between Florida and the Bahamas, extends across the Atlantic towards the western shores of Europe. He calculates that by this current alone an amount of heat equal to that received on the entire surface of the earth in a zone thirty-two miles in breadth on each side of the equator is carried from the tropics to the cooler regions of the northern hemisphere. Mr. Croll has, I think, victoriously replied to several of the objections opposed to this portion of his argument. His estimate of the volume of water transferred by the Gulf-stream from the tropics to the northern part of the Atlantic, which he reckons at the annual amount of about 166,000 cubic miles, is, I think, in no degree exaggerated; and I also think that he is warranted in estimating the mean initial temperature at about 65° Fahr. I am, however, persuaded that in assuming 40° Fahr. as the temperature to which, on an average, this vast body of water is reduced before it returns to the equatorial zone, Mr. Croll has gone beyond the probable limit. A large part of the stream is diverted eastward about the latitude of the Azores, and is never cooled much below 55° Fahr. before the waters enter the return current on the eastern side of the Atlantic basin; and I believe that, if we allow the water of the Gulf-stream to undergo an average loss of temperature of 20° Fahr., we shall be more likely to exaggerate than to underrate the amount of cooling.

In insisting on the importance of the Gulf-stream in modify

ing the climate of Europe and the adjacent parts of the arctic zone, Mr. Croll agrees with many preceding writers; but, so far as I know, he was the first to suggest that in consequence of the greater persistency of the south-east trade-winds, which ordinarily extend up to, and, at some seasons, even north of, the equator, the warm waters of the Northern Atlantic derive a large share of the heat which is carried to the temperate and arctic zones from the southern hemisphere. Applying the same reasoning to the currents of the Pacific Ocean, Mr. Croll arrives at the general conclusion ("Climate and Time," p. 94) that "the amount of heat transferred from the southern hemisphere to the northern is equal to all the heat falling within fifty-two miles on each side of the equator."

I do not believe that the facts on which Mr. Croll bases this essential portion of his theory are sufficiently established. With regard to the Atlantic, I have expressed in the text (p. 344) an opinion, derived from conversations with practical seamen, that in the Atlantic the trade-winds of the northern are stronger than those of the southern hemisphere. That opinion, I am disposed, on further examination, to regard as incorrect. I believe that the north-east trade-winds often blow with greater force; but, taking the average of the entire year, I now think there can be no doubt that the south-east trade-winds extend However this may

over a wider area in the equatorial zone. be, our knowledge of the currents of the Atlantic does not, I think, authorize us to conclude that the portion of heated water carried from the southern to the northern hemisphere is nearly so large as Mr. Croll has estimated. If the heat of the Gulfstream were mainly supplied, as Mr. Croll contends, from that source, there should be a marked difference in the volume and temperature of the current, between the season when the northeast trade-winds approach the equator and that in which the south-east trades prevail to the north of the line, for which there is no evidence.

As regards the currents and winds of the Pacific, in spite of one considerable exception, to which I shall further allude, I think that the balance of evidence points to a greater prevalence of the south-east trade-winds, and to the probable transference of some portion of the equatorial waters from the southern to the northern hemisphere.

For the present discussion it is best to accept Mr. Croll's estimate, and to compare the amount of heat which he supposes to be transferred from one hemisphere to the other with the total amount which is received annually from the sun on each hemisphere. For this purpose I have taken the known areas of the torrid, temperate, and frigid zones respectively, and, following Mr. Croll, I have adopted Mr. Meech's estimate of the average amount of heat, per unit of surface, received from the sun in each zone, irrespective of absorption by the atmosphere. To estimate the proportion of heat which actually reaches the surface, I have adopted Pouillet's measure of the proportion of solar radiation cut off at vertical incidence, which is 24 per cent. I assume 28 per cent. to be the average loss in the torrid zone, 50 per cent. in the temperate zone, and 75 per cent. in the frigid zone.* The resulting figures, showing the proportional amount of heat annually received on the surface of each zone, and on the entire hemisphere, are as follows:

Calculating, on the same basis, the amount received on a zone one mile wide at the equator, allowing a loss of 25 per cent. from atmospheric absorption, and multiplying the result by 104, I obtain the number 233'1, or rather more than one twentyfifth part of the entire heat annually received from the sun by each hemisphere.

To trace the results of such a transfer of heat from one hemisphere to the other, I shall adopt a mode of reasoning, sanctioned by the great authority of Sir John Herschel, to which Mr. Croll frequently resorts. It is by solar heat that the surface of the earth is raised above the temperature of space, which is assumed to be 239 degrees below the zero of Fahrenheit's scale. Adopting Ferrel's estimate, I take the mean temperature of the northern hemisphere at 59'5° Fahr., or 298 degrees above the temperature of space. To maintain this temperature, it

* Viewed in the light of Mr. Langley's recent researches on solar radiation, all these numerical determinations are probably far from the truth; but the errors do not much affect the present argument.

receives one-half of the amount of solar radiation which reaches the earth, and in addition, on Mr. Croll's hypothesis, one twentyfifth part of that which reaches the southern hemisphere. It follows that the heat available to raise the southern hemisphere above the temperature of space stands to that which is received by the northern hemisphere in the ratio of 24 : 26, and that the mean temperature of the southern hemisphere should be 298.5 × 1, or 275'5° above the temperature of space; so that, in ordinary language, the mean temperature of the southern hemisphere should be 36·5° Fahr. If the fact corresponded with this result of theory, it would not be necessary to invoke increased eccentricity of the earth's orbit to account for the extreme cold of one hemisphere, seeing that the actual conditions would suffice to completely alter their relative tempera

tures.

It occurs to me, however, that, on further consideration, Mr. Croll would reduce his estimate of the volume of heated water transferred from the southern to the northern hemisphere; but even if that estimate were reduced by one-half, we ought to find in the southern hemisphere a mean temperature of 47·8° Fahr., or nearly 12 degrees lower than that of our hemisphere.

We have already seen that, so far as climate depends on the relative position of the earth and the sun, we ought to find in the southern hemisphere climates of a more extreme character, with hotter summers and colder winters, than those to which we are accustomed. If it be true that through the agency of ocean-currents a considerable amount of heat is transferred to the northern hemisphere, that circumstance might serve to account for the fact that the summers of the southern are not generally hotter than those of the northern hemisphere; but it would, at the same time, tend to aggravate the severity of the southern winters.

At the time of the publication of Mr. Croll's earlier memoirs, there existed a general belief that the southern hemisphere was in fact notably cooler than our portion of the globe, and he naturally referred to the supposed fact as harmonizing with the general conclusions drawn by him from theory. But, imperfect as our knowledge of the southern hemisphere still is, a good deal of information has been obtained of late years. The only stations south of the fiftieth degree of latitude from which we

possess continuous observations are those mentioned in the text (p. 273); but we also know with sufficient accuracy the climates of two widely separated islands lying about 50° south ; and from these we derive results widely different from those to which we were led by theoretical considerations. The following table gives approximately the mean temperatures, on Fahrenheit's scale, for the year and for the hottest and coldest months of the places referred to in the southern hemisphere, and the means for corresponding latitudes in the northern hemisphere :

If we compare the mean results of these five stations with those for corresponding latitudes in the northern hemisphere, we find that the summers are cooler and the winters very much milder, and that in the latitudes between 50° and 55° the mean annual temperature is notably higher. In Kerguelen Land alone the mean annual temperature is lower than the normal for the same latitude north of the equator; but that island is evidently exposed to exceptional conditions.

* The observations at Stanley Harbour, which are those adopted by Dr. Hann (Klimatologie, p. 697), show temperatures notably lower than those recorded for a place in the islands lying farther south, which are given in the Zeitschrift der Esterreichischen Gesellschaft für Meteorologie, vol. v. p. 369. The mean of the two is probably nearly correct.

†These figures are derived from the tables given in the Anales de la Oficina Meteorologica Argentina, by B. Gould, vol. iii. The figures show a considerable amount of annual variation. The monthly means of the six months from February to July, 1879, exceed those of the same period in 1878 by more than 2° Fahr.