quite three times as great as the elevation of the highest of our mountains above the sea's level, but it is almost five times as great as that of the eastern plateau of Thibet.

We ought here to observe that the results of the earth's compression, which have been obtained by mere measurements of a degree, or by combinations of the former with pendulum measurements, show far less 24 considerable differences in the amount of the equinoctial bulging than we should have been disposed at first sight to conclude from the fractional numbers. The difference of the polar compressions (and) amounts to only about 7000 feet in the difference of the major and minor axes, basing the calculation on both extreme numerical limits; and this is not twice the elevation of the small mountains of the Brocken and of Vesuvius; the difference being only about one-tenth

24 It has often seemed to me as if the amount of the compression of the earth was regarded as somewhat doubtful merely from our wish to attain an unnecessary degree of accuracy. If we take the values of the compression at 1,300, šō, zš, we find that the difference of both radii is equal to 10,554, 10,905, 11,281, 11,684 toises, or 67,488, 69,554, 73,137, 74,714 feet. The fluctuation of 30 units in the denominator produces only a fluctuation of 1,130 toises or 7,126 feet in the polar radius, an amount which, when compared with the visible inequalities of the earth's surface appears so very inconsiderable, that I am often surprised to find that the experiments coincide within such closely approximating limits. Individual observations scattered over wide surfaces will indeed teach us little more than what we already know, but it would be of considerable importance to connect together all the measurements that have been made over the entire surface of Europe, including in this calculation all astronomically determined points." (Bessel, in a letter addressed to myself, December, 1828.) Even if this plan were carried out, we should then only know the form of that portion of the earth, which may be regarded as a peninsular projection, extending westward, about sixty-six and a half degrees from the great Asiatic Continent. The steppes of Northern Asia, even the middle Kirghis steppe, a considerable portion of which I have myself seen, are often interspersed with hills, and in respect to uninterrupted levels, cannot be compared with the Pampas of Buenos Ayres, or the Llanos of Venezuela. The latter, which are far removed from all mountain chains and consist immediately below the surface of secondary and tertiary strata, having a very uniform and low degree of density, might by differences in the results of pendulum vibrations, yield very decisive conclusions in reference to the local constitution of the deep internal strata of the earth. - Compare my Views of Nature, pp. 2-8, 29-32.

of the bulging which would be yielded by a polar compression of

As soon as it had been ascertained by more accurate measurements of a degree, made at very different latitudes, that the earth could not be uniformly dense in its interior, (because the results showed that the compression was very much less than had been assumed by Newton (), and much greater than was supposed by Huygens (), who considered that all forces of attraction were combined in the centre of the earth,) the connection between the amount of compression and the law of density in the interior of our earth necessarily became a very important object of analytical calculation. Theoretical speculations regarding gravity very early led to the consideration of the attraction of large mountain masses, which rise freely and precipitously into the atmosphere from the dried surface of our planet. ton, in his Treatise of the System of the World in a Popular Way, 1728, endeavoured to determine what amount of deviation from the perpendicular direction the pendulum would experience from a mountain 2,665 feet in height and 5,330 feet in diameter. This consideration very probably gave occasion to the unsatisfactory experiments, which were made by Bouguer on Chimborazo, by Maskelyne and

25

New

25 Bouguer who had been induced by La Condamine to institute experiments on the deviation of the plummet near the mountain of Chimborazo, does not allude in his Figure de la Terre, pp. 364-394 to Newton's proposition. Unfortunately the most skilful of the two travellers did not observe on the east and western sides of the colossal mountain, having limited his experiments (December, 1738) to two stations lying on the same side of Chimborazo, first in a southerly direction 61° 30′ West, about 4,572 toises or 29,326 feet from the centre of the mountain, and then to the South 16° West (distance 1,753 toises or 11,210 feet). The first of these stations lay in a district with which I am well acquainted, and probably at the same elevation as the small alpine lake of Yana-cocha, and the other in the pumice-stone plain of the Arenal (La Condamine, Voyage à l'Equateur, pp. 68-70). The deviation yielded by the altitudes of the stars, was, contrary to all expectation, only 7."5 which was ascribed by the observers themselves to the difficulty of making observations so immediately in the vicinity of the limit of perpetual snow, to the want of accuracy in their instruments, and above all to the great cavities which were conjectured to exist within this colossal trachytic mountain. I have already expressed many doubts, based upon geological grounds, as to this assumption of very large cavities, and of the very inconsiderable mass of the tra

Hutton on Shehallien, near Blair-Athol, in Perthshire; to the comparison of pendulum lengths on a plain lying at an elevation of 6000 feet and at the level of the sea (as for instance Carlini's observations at the Hospice of Mont Cenis, and Biot and Mathieu's at Bordeaux); and lastly to the delicate and thoroughly decisive experiments undertaken in 1837 by Reich and Bailey with the ingeniously constructed torsionbalance which was invented by John Mitchell and subsequently given to Cavendish by Wollaston.26 The three modes of determining the density of our planet (by vicinity to a mountain mass, elevation of a mountainous plateau, and the balance) have already been so circumstantially detailed in a former part of the Cosmos (vol. i, p. 158), that it only remains for us to notice the experiments given in Reich's new treatise, and prosecuted by that indefatigable observer during the interval between the years 1847 and 1850.27

chytic dome of Chimborazo. South-south-east of this mountain, near the Indian village of Calpi, lies the volcanic cone of Yana-urcu, which I carefully investigated in concert with Bonpland, and which is certainly of more recent origin then the elevation of the great domeshaped trachytic mountain, in which neither I nor Boussingault could discover anything analogous to a crater. See the Ascent of Chimborazo in my Kleine Schriften, Bd. i, s. 138.

26 Baily, Exper. with the Torsion Rod for determining the mean density of the earth, 1843, p. 6; John Herschel, Memoir of Francis Baily, 1845, p. 24.

27 Reich, Neue Versuche mit der Drehwage, in the Abhandl. der mathem. physischen Classe der Kön. Sächsischen Gesellschaft der Wissenschaften zu Leipzig, 1852, Bd. i, s. 405, 418. The most recent experiments of my respected friend Professor Reich, approximate somewhat more closely to the results given in Baily's admirable work. I have obtained the mean 5.5772 from the whole series of experiments: (a) with the tin ball and the longer thicker copper wire, the result was 5.5712, with a probable error of 0.0113; (b) with the tin ball, and with the shorter thinner copper wire, as well as with the tin ball and the bi-filar iron wire, 5.5832, with a probable error of 0.0149. Taking this error into account, the mean in (a) and (b) is 5.5756. The result obtained by Baily, and which was certainly deduced from a larger number of experiments (5.660), might indeed give us a somewhat higher density, as it obviously rose in proportion to the greater lightness of the balls that were used in the experiments, which were either of glass or ivory. (Reich in Poggend. Annalen, Bd. lxxxv, s. 190. Compare also Whitehead Hearn in the Philos. Transact. for 1847, pp. 217-229.) The motion of the torsion balance was observed by Baily by means of the reflection of a scale obtained from a mirror, which was attached to the middle of the

The whole may in accordance with the present state of our knowledge be arranged in the following manner :

Shehallien, according to the mean of the maximum 4.867 and the minimum 4.559, as found by Playfair...

4.713

Mont Cenis, observations of Carlini, with the cor

rection of Giulio,

4.950

The torsion-balance, Cavendish (according to Baily's

calculation)

5.448

Reich, 1838

5.440

Baily, 1842

5.660

Reich, 1847-1850

5.577

The mean of the two last results gives 5.62 for the density of the earth (taking that of water as 1), and consequently much more than the densest finely granular basalt, which according to the numerous experiments of Leonhard varies from 2.95 to 3.67, and more than that of magnetic iron (4,9 to 5.2), and not much less than that of the native arsenic of Marienberg or Joachimsthal. We have already elsewhere observed (Cosmos, vol, i, p. 159) that from the great distribution of secondary and tertiary formations, and of those upheaved strata which constitute the visible continental part of our earth's surface (the plutonic and volcanic upheavals being scattered in the form of islands over a small area of space), the solid portion of the upper part of the earth's crust possesses a density scarcely reaching from 2.4 to 2.6. If we assume with Rigaud that the relation of the solid to the fluid oceanic surface of our globe is as 10: 27, and if further we consider that the latter has been found by experiments with the sounding lead to extend to a depth of 27,700 feet, the whole density of the upper strata, which underlie the dry and oceanic surfaces, scarcely equals 1.5. The distinguished geometrician Plana has correctly observed that the author of the Mécanique Céleste was in error, when he ascribed to the upper stratum of the earth a density equal to that of granite,

balance, a method that had been first suggested by Reich, and was employed by Gauss in his magnetic observations. The use of such a mirror, which is of great importance from the exactness with which the scale may be read off, was proposed by Poggendorff as early as the year 1826 (Annalen der Physik. Bd. vii, s. 121).

which, moreover, he estimated somewhat highly at 3, which would give him 10.047 for the density of the centre of the earth.28 This density would, according to Plana, be 16.27 if we assume that of the upper strata = 1.83, which differs but slightly from the total density of 1.5 or 1.6 of the earth's crust. The vertical pendulum, no less than the horizontal torsion balance, may certainly be designated as a geognostic instrument; but the geology of the inaccessible parts of the interior of our globe is, like the astrognosy of the unilluminated celestial bodies, to be received with considerable cau

28 Laplace, Mécanique Céleste, éd. de 1846, t. v, p. 57. The mean specific weight of granite cannot be set down at more than 2.7, since the bi-axial white potash-mica, and green uni-axial magnesia-mica range from 2.85 to 3.1, whilst the other constituents of this rock, namely quartz and felspar are 2.56 and 2.65. Even oligoclase is only 2.68. If hornblende rises as high as 3.17, syenite, in which felspar always predominates, never rises above 2.8. As argillaceous schist varies from 2.69 to 2.78, while pure dolomite, lying below limestone, equals only 2.88, chalk 2.72, and gypsum and rocksalt only 2.3, I consider that the density of those continental parts of the crust of our earth, which are appreciable to us should be placed at 2.6 rather than at 2.4. Laplace, on the supposition that the earth's density increases in arithmetical progression from the surface towards the centre, and on the assumption (which is assuredly erroneous) that the density of the upper stratum is equal to 3, has found 4.7647 for the mean density of the whole earth, which deviates very considerably from the results obtained by Reich (5.577) and by Baily (5.660); this deviation being much greater than could be accounted for by the probable error of observation. In a recent discussion on the hypothesis of Laplace, which will soon form a very interesting paper in Schumacher's Astr. Nachrichten, Plana has arrived at the result that, by a different method of treating this hypothesis, Reich's mean density of the earth, and the density of the dry and oceanic superficial strata, which I estimated at 1.6, as well as the ellipticity, within the limits that seem probable for the latter value, may be very closely approximated to. "If the compressibility of the substances of which the earth is formed," writes the Turin geometrician, "has given rise to regular strata, nearly elliptical in form, and having a density which increases from the surface towards the centre, we may be allowed to suppose that these strata, in the act of becoming consolidated, have experienced modifications, which, although they are actually very small, are nevertheless large enough to preclude the possibility of our deducing, with all the precision that we could desire, the condition of the solid earth from its prior state of fluidity. This reflection has made me attach the greater weight to the first hypothesis advanced by the author of the Mécanique Céleste, and I have consequently determined upon submitting it to a new investigation."

VOL. V.

D