meteor is vertical. We have very strong evidence, showing that 70 miles is about the height at which meteors appear, the evidence of meteors appearing at a greater height being very doubtful. Hence, when a meteor is seen low down towards the horizon, it may be confidently assumed that the point over which this meteor is vertical lies within 750 miles of the place of observation. Now the ovals and circle in Maps 1 and 2 mark the limits of the space over some point of which a meteor must be vertical to be seen from the centre of the space. For instance, a meteor appearing at a point vertical over Madrid, or Turin, or Berlin, or Stockholm, might just be visible from London, appearing just above the horizon; but a meteor vertical over Gibraltar, or Rome, or St. Petersburg, would not be visible in England.

Now, if we consider Map 1, we shall see that about two hours before the time indicated by that map (a quarter past twelve at night), London is just becoming visible on the edge of the earth's disk; but the edge of the oval space round London comes into view more than an hour earlier that is, at about nine o'clock. This is the earliest hour at which a member of the November system can by any possibility be seen in London. Meteors seen at this hour would be momentarily visible in the eastern horizon, moving upwards. When London comes to the border of the visible hemisphere, meteors may be looked for over the whole space between the eastern horizon (that is from south, through east to north) and the zenith, travelling (more or less) upwards unless they appeared nearly towards the north or south, when their motion would be horizontal. When the whole of the London oval space is in view, meteors may be looked for over the whole heavens. A little consideration will show that at and after this time, conspicuous meteors will be seen more plentifully over the western half of the heavens. If the mere number of meteors indeed, were alone considered, the contrary would be the case. But the paths of meteors being from a point east of London (it is clear that both in Map 1 and in Map 2, we are looking at London from the east), they would have in general an apparently westward motion, and all those having long visible tracks would be towards the west.

It is also evident from Figs. 1 and 2, that meteors increase in number (cæteris paribus) as England, through the earth's rotation, approaches the centre of the disk visible from the radiant point, or-which amounts to the same thing-as the radiant point rises above the horizon. It is clear, for instance, that the oval space round England in Fig. 2, is greater than the oval in Fig. 1; and that at an hour later than that indicated in Fig. 2, the oval is yet greater. The oval round England is

greatest at about a quarter past six, when the meridian of London is a diameter of the disk. The effects due to this cause of variation ought to be considered in estimating the actual changes in the richness of the shooting star-stream as the earth traverses different strata. For instance, the increase which actually occurred after midnight, last November, was partly due to this cause, while the diminution which took place subsequently to 1h. 30m. or 1h. 45m., was partly checked by this cause.

Let us stay for a moment to compare with the effects just considered, those occurring in other latitudes. It is clear from Figs. 1 and 2, that countries in northern latitudes are more favourably situated than countries in southern latitudes, as respects their chance of seeing the November star-shower. Thus, if we consider the short part of the arc traversed by Cape Town, which lies within the darkened part of the disk, it is clear that the á priori probability that observers there will see the phenomenon is small. The hour at which Cape Town reaches the diametral meridian being about 6h. 15m., Cape time, it is clear that the moment at which Cape Town enters on the part of the disk visible from the radiant point, is about 2h. 15m. The oval round Cape Town begins to enter this part of the disk rather more than one hour earlier. Thus, unless the phenomenon occurs between about one o'clock and day-break (it will be seen that Cape Town enters the enlightened half disk, or, in other words, the sun rises there soon after five), it will not be seen at all at Cape Town; and that it should be well. seen, it is necessary that the epoch of maximum richness should occur between 1h. 30m. and 3h. 30m. Cape time. It happened last November, that the shower reached its maximum at 2h. A.M., Cape time, and was, therefore, well seen there.*

In tropical regions north of the equator, which enter on the hemisphere turned towards the radiant during the continuance of the shower, the display is likely to be grander than elsewhere, since the circular space around any point in such regions would be seen as an oval of much less eccentricity than that round places in high latitudes, during a part at

For the same reason that meteors are more commonly seen in northern latitudes from July to December, they are more commonly seen in southern latitudes from January to June. An examination of Figs. 1 and 2 will illustrate the cause of this peculiarity, viz.:—the presentation of the northern and southern poles respectively towards the direction of the earth's motion. It is worthy of notice that Mr. Maclear records the observation of several meteors last November, before the hour at which Cape Town (or the space included within the oval in Fig. 2) entered on the hemisphere turned towards the radiant; or, in other words, before the radiant rose above the horizon: but none of these belong to the November system, as is evidenced by the direction of their motion.

least, of their passage across the darkened part of the disk. At Calcutta, for instance, the boundary of visibility is appreciably circular (as shown in Fig. 1) a short time before sunrise. At this hour, last November, the shower had not reached its full splendour, and therefore, the richer part of the display was not seen in Calcutta. In Nubia, Egypt, Asia Minor, and Greece, the shower was more favourably seen. Mr. Schmidt, for instance, reported a very rich display at Athens, reaching its maximum at 2h. 15m. local time, or about 12h. 45m. Greenwich time; very nearly the hour illustrated in Fig. 1. The display in India (at Kishnagur, fifty miles from Calcutta) began before four o'clock, and continued till daylight. At 4h. A.M., Calcutta mean time, which corresponds to 10 P.M., Greenwich time, London had not yet reached a position for a favourable view of the display.

It will be seen from Figs. 1 and 2, that during nearly the whole time that the display continued, last year, in England, every visible shooting star was travelling towards the earth's surface, not grazing the atmosphere. Thus no shooting star which fell within the oval line marked round England in Fig. 1, or in Fig. 2, could have failed to reach the earth's surface, unless dissipated in the upper regions of air. And, indeed, independently of the consideration of the November shower, and its radiant, it is quite clear that of meteors which pass into an atmosphere, by far the larger number travel in a line which produced meets the earth's solid surface. For, in whatever direction a meteor stream is travelling, the earth, seen from the radiant point of the stream, must present an appearance corresponding to that illustrated in Figs. 1 and 2. The pole may be more or less bowed towards, or from, the direction in which the meteors are travelling (relatively) towards the earth, and other countries than those presented in the figures, may be turned towards the meteor-flight; but a circular disk, apparently fringed with a comparatively narrow border of atmosphere, must in every case be presented towards the meteor-stream. Only those meteors which impinge on this fringe, a circular ring 70 miles wide,* can possibly free themselves by passing through (or grazing) the atmospheric envelope. All those meteors which are making for the apparent disk of the solid earth, a circle, nearly 4,000 miles in radius, must inevitably reach the earth, either in a solid form or in the form of meteoric dust, after being dissipated in their passage through the upper atmospheric layers. Assuming that every meteor making for the fringe escapes, which is,

Of course, not in reality such a ring, but apparently so viewed from the radiant point of the meteor-flight; and intercepting the same proportion of meteors as if actually so.

however, utterly improbable, it may easily be calculated that for every meteor grazing our atmosphere (at a height not exceeding 70 miles), twenty-eight travel directly towards the earth's surface. But the proportion must in reality be very much greater, since our supposition implies the possibility of a meteor travelling through the air in a direction actually tangent to the earth's surface, or passing through about 1,450 miles of air, including the densest strata. Since meteors seldom penetrate to a vertical depth of more than twenty or thirty miles, without dissolution, it is very unlikely that meteors travelling parallel to the horizon should penetrate to a vertical depth even of ten or fifteen miles-since, to do so, their actual path through the air would be many times longer. Assuming that meteors could escape after penetrating in this manner to a depth of twenty miles, we should have, for every meteor so escaping, almost exactly one hundred whose substance, whole or dissolved, would reach the earth. Even escaping meteors would never again appear as members of the November shower, since their orbit, after grazing contact of the kind supposed, would be very different (owing chiefly to their loss of velocity) from that they originally pursued.

In the fact that such multitudes of meteors have, during so many and such brilliant displays of November showers as have been recorded, been stolen by the earth from the stream to which they belonged, serves to afford some conception of the immense number of meteors forming the November stream. Yet clearer views will be formed on this point if we consider the evidence we have respecting the length, breadth, and thickness of the cluster, during the passage through which the display is visible. I have not space to dwell here on Adams' investigation of the meteoric orbit. But it is necessary to point out that we must now greatly increase our estimate of the length of the cluster causing the November showers. The recurrence of displays during two or three consecutive years was simply accounted for on the theory of a nearly circular orbit, without assuming for the cluster a length of more than a few millions of miles. Now that we know that the meteorflight travels in an orbit of great eccentricity, and with a period of 33 years, we know that the portion passed through by the earth in one year is several hundreds of millions of miles away, when the earth next passes through the meteor orbit. Hence the recurrence of displays leads us to estimate the length of the cluster by hundreds of millions of miles, instead of by mere millions.

Next, for the breadth of the stream. On this point we have no exact information. It is sometimes assumed that the fact that the display may be seen in one hemisphere, while in

another it is not seen (as last year, for instance, in America), points to a limit of breadth. But this is not the case. If we consider Figs. 1 and 2 we shall see that America was on the sheltered side of the earth during the whole time of the display. When America had come to the side turned towards the radiant, the earth's globe had, in all probability, passed through the meteor-stream. So that the limits of the thickness, and not of the breadth of the stream, were indicated by the non-visibility of the meteors in America. Before the display had begun in England, the meteors, were seen from Kishnagur, fifty miles north of Calcutta, and they continued visible until the time of sunrise there. This would assign a breadth of not less than 4,000 miles to the stream. But as, throughout the continuance of the display, the earth was crossing the breadth of the stream at the rate of about 1,000 miles an hour, we can assert positively that the breadth of the stream exceeded 6,000 miles. In reality, however, a very much greater breadth may be assigned, with great probability, to the meteor-stream. For if we consider the nature of the stream and the manner in which it has been probably generated in the track of Comet I., 1866, we shall see the great probability that its breadth exceeds its thickness. For the causes tending to make meteors leave the mean plane of motion, are much less efficient than those tending to distribute the meteors over that plane. Now the earth, during the time of the display, was crossing the thickness of the meteor-stream at the rate of about 18,000 miles an hour. Therefore, since the display lasted at least six hours (counting from the time of its being observed in India, when England was, as yet, on the earth's sheltered side), we cannot assign to the stream a less thickness than 100,000 miles. The breadth is probably at least ten times as great.

It may be assumed as certain, that it is the passage of the earth through the thickness of the meteor-stream which limits the duration of the display.

I shall conclude by quoting two observations, showing that the fine powder in which meteors reach the earth may be detected. Dr. Reichenbach collected dust from the top of a high mountain, which had never been touched by spade or pickaxe; and on analysis he found this dust to consist of almost identically the same elements as those of which meteoric-stones are composed-nickel, cobalt, iron, and phosphorus. Again, Dr. Phipson notes, that "when a glass, covered with pure glycerine, is exposed to a strong wind, late in November, it receives a certain number of black angular particles," which "can be dissolved in strong hydrochloric acid, and produce yellow chloride of iron upon the glass-plate." I quote these