Dr. Carpenter illustrated this theory, or rather the combined action of polar cold and equatorial heat, by an experiment, the plan of which had occurred also to myself, and been described by me in conversation somewhat earlier. 'A long narrow trough having glass sides was filled with water, and a piece of ice was wedged in at one end between its side plates just beneath the top, whilst the surface of the water at the other end was warmed by a piece of metal, of which a part projected beyond the trough, and was heated by a spirit lamp placed beneath it; thus representing the relative thermal conditions of the polar and equatorial basins. A colouring liquid viscid enough to hold together in the water, while mixing with it sufficiently to move as its moves, being then introduced, the liquid as it impinged on the ice was seen to sink rapidly to the bottom, then to flow slowly along the floor of the trough towards the opposite extremity, then gradually to rise beneath the heated plate, and then to flow slowly along the surface towards the glacial end, repeating the same movement until the ice had melted.'

' Fresh water expands with reduction of temperature, near the freezing point, and hence, becoming lighter, the descending motion above described is interfered with in the case of fresh water.

It will be observed that in this experiment the effect of cold is not exerted alone, so that it by no means proves that the arctic cold is the chief agent in producing the system of oceanic circulation. Moreover, the conditions of the polar and equatorial basins are in one respect not accurately (or even nearly) reproduced, for the real arctic area is very much smaller, compared with the real equatorial area, than in the case of the experiment. Indeed it appears to me that Dr. Carpenter paid far too little attention to the relative smallness of the arctic area. This may have been partly due to the erroneous ideas suggested by the ordinary maps on Mercator's Projection, in which, as I have already mentioned, the arctic regions are enormously exaggerated. It is almost impossible to study such a map as that which illustrates this paper (see page 217) without feeling that the theory presented by Dr. Carpenter will scarcely hold water, or rather-if this way of presenting the argument be permitted-that the arctic area does not hold water enough to produce the effects described by Dr. Carpenter. For in that map the whole area of the Arctic Ocean is presented;l and from out of that area, be it noted, must come the northern supply of descending water, not only for the Atlantic equatorial current, but for the much larger equatorial current of the Pacific, if Dr. Carpenter's theory be sound.

The following letter, written by Sir John Herschel only a few weeks before his lamented decease, has been very widely quoted in favour of Dr. Carpenter's theory; yet if carefully studied it will be found rather to set forth the strength of the theory advocated a year earlier by the present writer. In this letter, at least, Sir John Herschel appears to be disposed, in so far as he concedes the efficiency of heat, cold, and evaporation, to incline to the equatorial action as the most important. Answering Dr. Carpenter, who had addressed a letter to him on the subject, he says: 'After well considering all you say, as well as the common-sense of the matter, and the experience of our hot-water circulation pipes in our green-houses, etc, there is no refusing to admit that an oceanic circulation of some sort must arise from mere heat, cold, and evaporation, as verae causae; and you have brought forward with singular emphasis 2 the more powerful action of the polar cold, or rather, the more intense action, as its maximum effect is limited to a much smaller area than that of the maximum of equatorial heat. The action of the trade and countertrade winds, in like manner, cannot be ignored; and henceforward the question of ocean-currents will have to be considered under a twofold point of view.'

1 The bounding lines drawn from the pole on the right and left of the white space represent one and the same meridian.

2 In Sir John Herschel's letters one can often recognise slight touches we will not say of sarcasm (for he was incapable of saying aught that could be considered bitter or unpleasant), but of what may be described as a humorous suggestiveness.

It appears to me that not only is the equatorial or rather tropical action much wider in range, but it is also more intense than the polar action. For, let us consider what happens during the heat of the day over the tropical Atlantic. Here, over an area enormously exceeding the whole arctic basin (we are considering, be it understood, only the northern part of the system of circulation) a process of evaporation is taking place at so rapid a rate as to furnish almost the whole of that rain-supply whence the rivers of Europe and North America (east of the Rocky Mountains) take their origin. There is thus produced a continual defect of pressure, not merely along an equatorial strip, but so far as 20 or even 30 degrees of north latitude, while the downfall of rain which, taking one part with another of the temperate and sub-arctic Atlantic, may be regarded as incessant, continually adds to the pressure in these last-mentioned regions. That on the whole there must result a most effective excess of pressure over the temperate zone of the Atlantic, as compared with the tropical and equatorial portion, seems to me indisputable. Now, if we compare this with the effects of refrigeration over the relatively insignificant arctic area, which as I have said has to supply the North

Pacific submarine circulation (if Dr. Carpenter's theory be true), as well as that of the North Atlantic, we can scarcely doubt, as it seems to me, which cause is the more effective. I would venture to predict that if Dr. Carpenter's experiment were tried first with the ice alone to produce circulation, and secondly with the heat alone, the superior efficiency of the latter cause would be at once recognised; but I much more confidently predict that if, as in the experiment I myself proposed, the relative areas of the equatorial and arctic basins were represented, there would be found to be scarcely any comparison between the effects of arctic cold and equatorial heat, so largely would the latter predominate.