|

d'Arc, Histoire des Conquêtes des Normands en Italie, en Sicile, et en Grèce.)

a more musical strain, resembling somewhat, in the mellowness of its tones, the song of the fifing Baltimore. The syllables to which I have hearkened appear like 'tshoove 'wait 'wait, 'vehowit wait, and 'wait, 'vehowit vea wait, with other additions of harmony, for which no words are adequate. This pleasing and highly musical meandering ditty is delivered for hours, in a contemplative mood, in the same tree with his busy consort. If surprised, they flit together, but soon return to their favourite station in the spreading boughs of the shady oak or hickory. This song has some resemblance to that of the Red-eyed Vireo in its compass and strain, though much superior, the 'wait 'wait being whistled very sweetly in several tones, and with emphasis; so that, upon the whole, our Pyranga may be considered as duly entitled to various excellencies, being harmless to the farmer, brilliant in plumage, and harmo-Tancred, in Syria and Palestine, have been immortalized nious in voice.'

Nest, Food, &c.-The same author describes the nest (which is built about the middle of May, on the horizontal branch of some shady forest-tree, commonly an oak, but sometimes in an orchard tree) as but slightly put together, and usually framed of broken rigid stalks of dry weeds or slender fir-twigs, loosely interlaced together, and partly tied with narrow strips of Indian hemp (Apocynum), some slender grass-leaves, and pea-vine runners (Amphicarpa), or other frail materials; the interior being sometimes lined with the slender, wiry, brown stalks of the Canadian cistus (Helianthemum), or with slender pine-leaves; the whole so thinly_platted as to admit the light through the interstices. The three or four eggs are dull blue, spotted with two or three shades of brown or purple, most numerous towards the larger end. As soon as their single brood, which is fledged early in July, is reared, they leave for the south, generally about the middle or end of August.

The female,' says this interesting author in continuation, 'shows great solicitude for the safety of her only brood; and, on an approach to the nest, appears to be in great distress and apprehension. When they are released from her more immediate protection, the male, at first cautious and distant, now attends and feeds them with activity, being altogether indifferent to that concealment which his gaudy dress seems to require from his natural enemies. attached to his now interesting brood is the Scarlet Tanager, that he has been known, at all hazards, to follow for half a mile one of his young, submitting to feed it attentively through the bars of a cage, and, with a devotion which despair could not damp, roost by it in the branches of the same tree with its prison.'

So

The food of this species consists mostly of winged insects, such as wasps, hornets, and wild bees, the smaller kind of beetles, and other Coleoptera. Seeds are supposed to be sometimes resorted to, and they are very fond of

whortle and other berries.

It is in August that the moult of the male, when he exchanges his nuptial scarlet for the greenish-yellow livery of the female,' commences. (Manual of the Ornithology of the United States and of Canada.)

TANAGRI'NÆ. [TANAGERS.]
TA'NAIS. [DON.]
TANARO. [Po.]

TANCRED, of Hauteville in Normandy, was a feudal paron who lived in the latter part of the tenth and beginning of the eleventh century. After doing military service for some years under Richard the Good, duke of Normandy, he retired to his hereditary mansion, where he lived poor, and reared up a numerous family of twelve sons and three daughters. All his sons were remarkable for their comeliness, their great strength, and their courage. The eldest, Serlon, followed William the Bastard in his conquest of England, and the others went successively to seek their fortune in Apulia, where Rainulf, another Norman adventurer, had already obtained the countship of Aversa from Sergius, duke of Naples. William, one of Tancred's sons, called Fier à bras,' or strong of arm, became count of Apulia, and after his death, his brother Robert, called Wiskard, or the wise,' became duke of Apulia and Calabria, and the founder of the Norman dynasty of Sicily. [SICILIES, TWO, History of.] Their father Tancred died at a very great age at Hauteville. Traces of the château of Tancred, according to old popular tradition, were still seen a few years since in a pretty valley near Hauteville, four miles north of the town of Marigny, in the arrondissement of Coutances department of La Manche. (Gaultier

TANCRED, son of Eudes, a Norman baron, and of Emma, sister of Robert Wiskard, duke of Apulia, according to some (Gaultier d'Arc, Histoire des Conquêtes des Normands en Italie, en Sicile, &c.), and nephew of Bohemund, son of Wiskard, and prince of Tarentum according to others (Giannone and the authorities he quotes), was serving with Bohemund under Roger, duke of Apulia, son and successor of Wiskard, at the siege of Amalfi, A.D. 1096, when the report of the great crusade which was preparing for the East determined Bohemund, who was not on good terms with Duke Roger, to join the Crusaders. Tancred followed him with a vast number of men from Apulia and Calabria. The exploits, true or fabulous, of by Tasso in his poem of La Gerusalemme.'

TANCRED, king of Sicily. [SICILIES, Two, History of.) TANGENT. In the article CONTACT We have given the first notion on this subject, which we now resume in a somewhat more general manner, annexing the usual details of formulæ, but without proof.

It is usual to apply the word tangent to the tangent straight line only, on which see DIRECTION: generalizing the definition, it will be as follows:-Of all curves of a given species, or contained under one equation, that one (B) is the tangent to a given curve (A) at a given point, which passes through that given point, and is nearest to the curve (A): meaning that no curve of the given species can pass through the given point, so as to pass between (B) and (A), immediately after leaving the point at which the two latter intersect.

To ascertain the degree of contact of two curves which meet in a point, proceed as follows. Let y = px and y=x be the equations of the curves, and a the abscissa at the point of contact; so that daya. At the point whose abscissa is a+h, the difference of the ordinates of the curves is, by Taylor's theorem,

m

h2

m+1

ha

....

(p'a-'a) h + (p′′a−4′′a) 2 + (p""a−4′′a)2.3+
as to which, generally speaking, it will be found that h
can be taken so small that the series shall be convergent:
if this be not so, the method of arresting the series given
in TAYLOR'S THEOREM must be employed. Now of two
series of the form Ah+Bh +.... the value of that in
which m is the greater will diminish without limit as com-
pared with the other, when h diminishes without limit.
Consequently, every curve y=x, which has 'a d'a, will
approach, before the point of contact is attained, nearer to
y=pr than any other in which 'a is not p'a. Again,
when p'a 'a, those cases of y=4x in which "a=p"a,
will approach nearer to y=pr than any in which p'a is
closest possible contact with y=p when x=a;—give such
not "a; and so on. Hence, to make y=4x have the
values to the constants in y=4 as will satisfy as many as
possible of the equations paya, p'a=4'a, pa="a, &c.
consecutively from the beginning. This is a brief sketch,
which can be filled up from any elementary work; and the
following are the principal results:-

1. When the string of equations is satisfied up to
(n) (n)
paa, the contact is said to be of the nth order.
2. In contact of the nth order, the deflection (a+h)—
, and vanishes in a finite ratio
(a+h) diminishes with h

to it.

"+1

3. In contact of an even order, the curves intersect at

the point of contact; in contact of an odd order, they do not intersect at that point.

4. When curves have a contact of the nth order, no curve, having with either a contact of an order inferior to the nth at the same point, can pass between the two.

5. A straight line, generally speaking, can have only a contact of the first order with a curve; and the equation to the tangent straight line of the curve y=pr, when x=a, is y—pa=p'a(x-a). But if it should happen that

(n)

"a 0, "a=0, &c., up to 4a=0, then for that point the tangent has a contact of the nth order. Thus, at a point of contrary flexure the tangent has a contact of the second order, at least, with the curve.

6. A circle, generally speaking, can be made to have a contact of the second order with a curve, and the equation

-2

This circle cuts the curve, generally speaking: if not, as for example, at the vertices of an ellipse, it is evidence that the circle has a contact of some higher and odd order. The centre of the circle of curvature is a point on the normal, being that at which the normal touches the evolute. [INVOLUTE AND EVOLUTE.]

Not only is the term tangent most generally applied to the closest straight line only, but frequently only to that portion of the straight line which falls between the point of contact and the axis of x. Again, the normal is a straight line perpendicular to the tangent, drawn through the point of contact: but this term also is frequently applied only to that portion which falls between the point of contact and the axis of r. It is with reference to this limitation that the terms subtangent and subnormal are to be understood: the first meaning the distance from the foot of the tangent to the foot of the ordinate; the second that from the foot of the ordinate to that of the normal. The formula for the subtangent is pa÷p'u; that for the subnormal pa× p'a.

Let B be the angle made by the tangent with the axis of; usually the angle made by that part of the tangent which has positive ordinates with the positive side of the Then 3, at the point whose abscissa is x, is determined by the equation dy

axis of x.

dx

subnormal

dy tan ẞ= -; and subtangent y x dx 'Jy' If we take the more general mode of measurement proposed in SIGN, this equation remains equally true. Now, keeping strictly to that mode, let ẞ be the angle made by the tangent with the axis of x, the angle made by the radius vector with the axis of x, and μ that made by the tangent with the radius vector. It will be found, then, that in all cases

d Ꮎ

μ=ß-0, tan μ = "dr

Unless the mode of attributing signs be carefully attended to, these last equations, though always considered as universally true, are not so in reality.

2

We now come to the consideration of a surface. The mode of defining contact of a given order resembles that adopted with reference to a curve. Thus if = = (x, y) and = (x, y) be the equations of two surfaces coinciding when xa, yb, so that (a, b) = ↓ (a, b), then if the point be taken at which x=a+h, y=b+k, the contact of the two surfaces is of the nth order, when the deflection

$(a+h, b+k) −4 (a+h, b+k) being developed in powers of h and k by Taylor's Theorem, shows no terms lower than those of the form Ah" +Bh"-1k+. +Mk". This is tantamount to the following: two surfaces have a contact of the nth order when any plane whatever drawn through the point of contact cuts the surfaces in two curves which have a contact of the nth or a higher order.

...

Every surface has at every point a plane which has a complete contact of the first order. If z = p(x, y), and x, y, z be the co-ordinates of the point of contact, and ,, those of any point in the tangent plane, then the equation of the tangent plane is

dz

८

121 (−x)+
dx

dz dy

(n-y).

|

dx (?−z) = 0, n−y+

In the latter case, they are

dx2 dys dxdy , say U, at the point of contact. Imagine a plane to pass through the normal, cutting the surface in the curve (C) and the tangent plane in the straight line (L). Then, while the plane revolves about the normal, (L) is always tangent to (C).

1. Let U be positive. Then (L) has never more than a contact of the first order with (C), the surface nowhere passes through the tangent plane, and we have only such contact as is seen at any point of a sphere or ellipsoid.

2. Let U=0. Then (L) has never more than a contact of the first order with (C), except when the plane is in one position, in which there is a contact of a higher order. If U-0 at the point of contact only, and begin to take value at all adjacent points, nothing more would appear than in the last case, except that in one particular direction from the point of contact, and in its opposite, the surface would seem to grow nearer to the tangent plane than in any others. But if U=0 at all points of this surface, this approach to the tangent plane in one particular direction becomes more marked: for the surface lies on that plane in a straight line, that is to say, every tangent plane meets the surface in a straight line infinitely extended both ways; and the plane is tangent to the surface at every point of that straight line. Such surfaces, namely those in which U is always =0, are developable, or can be unrolled without any overlapping, rumpling, or tearing. Cones and cylinders are instances. Again, if U=0, not throughout the whole surface, but throughout one parits plane will be tangent to the surface at every point in ticular line upon it, that line will be a plane curve, and which it meets the surface.

3. Let U be negative. Then (L) has never more than a contact of the first order with (C), except in two different positions, in both of which there is contact of a higher order. Draw lines marking out these two positions of (L), and consequently dividing the tangent plane into four parts, with four angles round the point of contact. In one pair of the opposite angles, the surface lies on one side of the tangent plane, and in the other on the other.

Again, as the plane which revolves round the normal takes its different positions, the curvature of the section (C) changes. The two positions of the revolving plane in which the curvatures are greatest and least (algebraically) into the mathematical formulæ connected with this subare at right angles to one another. We shall not enter ject, but shall only endeavour to give a popular illustration of this remarkable point.

But if the equation be given in the form ø (x, y, z) = 0, will be at right angles to that of least rapid descent. The

make a tangent plane horizontal, there is absolutely no descent in one direction, or, by going along the tangent plane, we can remain entirely on the surface, in one certain direction, as before observed. And the direction of most rapid descent is at right angles to this direction of no descent.

To put a case of the third kind, suppose a saddle placed on a horse, and we take the lowest point of the seat. The tangent plane then cuts through the saddle horizontally. In some directions there is descent, in others ascent, with two directions in which there is, comparatively speaking. neither ascent nor descent. The direction of most rapid ascent, which is from the lowest point of the seat directly towards the head or tail of the animal, is at right angles to the direction of most rapid descent. Mathematically Speaking, the curvatures of the vertical sections are sometimes positive, and sometimes negative, and the direction of the greatest negative (or algebraically least) curvature is at right angles to the direction of the greatest positive (or algebraically greatest) curvature.

107° E. long. To the south of it is Tibet; to the west Chinese Turkistan, or the government of Thian-Shan Nanlu; and to the north Mongolia, of which also a portion is included within the lately erected province of Kansi. As the boundary-lines of the country are not politically determined, it is not possible to give an estimate of the area.

The southern portion of Tangut, or that which lies south of 38° N. lat., is one of the most mountainous tracts on the globe, and extends over the upper course of the river Hoang-ho and the basin of the lake of Khookhoo-nor. Along its southern border there is a very elevated range, which divides the upper courses of the rivers Hoang-ho and Yan-tse-kiang, and is called the Bayan Khara range. [BAYAN KHARA MOUNTAINS.] Another elevated range traverses the country in the same direction from east to west near 38° N. lat. This range rises at a short distance from the banks of the Hoang-ho north of the town of Lantcheou, and in its eastern part is called Kilian Shan; but farther west it takes the name of Nan Shan (or Southern Chain). It rises to a great elevation, especially towards the

As to points connected with the apparent physical cha-west, where many of their summits are covered with snow and racter of the tangent, which have been in various places referred to this article, it will be more convenient to consider them under the word VELOCITY.

TANGHI'NIA, the name of a genus of plants belonging to the natural order Apocynaceæ. This name was given by Aubert du Petit Thouars to the plant which produces the celebrated Tanghin poison of Madagascar. The genus possesses an infundibuliform corolla, with a clavate tube, and 5-toothed throat: the anthers are subsessile; the fruit is a drupe, with a fibrous ligneous putamen or stone, which contains one or two seeds. The specific name T. venenifera was given to the plant which yields the poison. It has dense leaves, with erect branches, and paniculated terminal flowers. At the time Du Petit Thouars described this plant, he stated that it was closely allied to the Cerbera Manghas; and since its cultivation by Mr. Telfair in the Mauritius, there can be no doubt of its belonging to the genus Cerbera, and the plant is now called C. Tanghin. In its native island this plant attains the size of a tree, and has a hard wood which may be used for many kinds of carpentry. But the part which yields the poison is the kernel of the fruit. Although this kernel is small, not much larger than an almond, Mr. Telfair says that it contains enough poison to kill twenty persons. Its great use in Madagascar was as a means of trial, the innocent being supposed able to resist its action, whilst the guilty suffered under its influence. Radama, the late king of Madagascar, was desirous of abolishing its use, but found great difficulty in doing so on account of the prejudices of the natives. Mr. Telfair witnessed a sad instance of its use. The king Radama was taken ill, and got well by the use of mercury; but this medicine affected his mouth, so that the impression produced upon his skid,' or physician, was that the king had been poisoned. He therefore insisted that the Tanghin should be administered to himself and all the servants of the household, in order to ascertain the guilty party. The king protested against the procedure, but in vain. The whole household were shut up during the night without food, and in the morning were brought out for trial. The presiding 'skid,' or physician, then pounded the Tanghin bean to a pulp between two stones, and applied a small quantity to the back of the tongue of each individual. The effects varied in different individuals. In some it produced vomiting, and the poison being ejected from the stomach, they recovered. In others convulsions were brought on with violent efforts at vomiting, which soon destroyed life. (Botanical Magazine, fol.

2968.)

TANGIER. [MAROCCO.] TANGLE. [SEA-WEEDS.] TANGUT is the historical name of a country in Asia, which occupies the centre of the eastern, more extensive, and more elevated table-land of that continent [ASIA, vol. i., p. 464], where a nation, which originally inhabited Tibet, and was called Tang, founded an empire in the seventh century, which was very powerful for a long time, and was overthrown by Genghis Khan in 1227. The country still goes by the name of Tangut, though at present a part of it is incorporated in the Chinese province of Kansi, whilst another is mostly in possession of two Mongol nations, the Olöth Tshoros and the Torbod Mongols.

Tangut borders on China Proper on the north-west, extending between 33° and 42° N. lat., and between 94° and

united by extensive glaciers. This mountain-chain is supposed to be connected with the Kuenluen range near 92° E. Jong. These two ranges above mentioned occupy a great portion of the country between 33° and 38° N. lat., and nearly the whole of the remainder of the country is filled up by a third range, which connects these two ranges, and extends from south-east to north-west, being on the north united to the Nan Shan, and on the south to the Bayan Khara Mountians. This chain bears the name of Sine Shan, or Snowy range, on account of the numerous summits which rise above the snow-line. The river Hoang-ho breaks through this range, but the huge rocky masses compel the river to make a great bend towards the west between 34° and 36° N. lat., and the circuit which the river makes shows the immense extent of these masses of rock. In this part of its course the river is said to be hemmed in by lofty mountains, so that no communication can be established along the banks. Its course above this bend is very imperfectly known, and the fabulous accounts of its sources show that they have never been visited even by Chinese geographers. The river enters a wide valley by a narrow gorge formed by two very elevated mountains a little above the town of Ho-cheou (36° N. lat. and 102° E. long.). At the opening of this gorge is a fortress, called Tsy-shy-kuan.

Tangut is separated from China Proper by a fourth range, the mountains of Sitan, which run south and north, being connected at their southern extremity with the Bayan Khara Mountains and the Siue Shan by an extensive mountain-knot, which is in the country formerly called Sifan, whence the chain has obtained its name. Though this range is less elevated than the Siue Shan, it rises in several places above the snow-line, and occupies a considerable width. It is supposed to terminate near the banks of the Hoang-ho, a few miles south of 38° N. lat. Opposite to it and on the northern banks of the river rises another chain, which may be considered as the continuation of the mountains of Šifan; but this range, which continues along the western bank of the river as far north as 42° N. lat., rises only to a moderate elevation, and is stated to occupy in many places only three or four miles in width: it is called Holang Shan, and slopes on the west down into the steppe of the Olöth Tshoros. This range is distinguished from all the other ranges of Tangut by being thickly wooded on its eastern declivity.

Only a small portion of the countries enclosed by these mountain masses is fit for cultivation. It does not appear that there is any cultivation in the upper valley of the Hoang-ho above the fortress of Tsy-shy-kuan. Below that place and as far as Lan-tcheou, the valley is wider, and narrow tracts along the banks of the river are cultivated and fertile. This part of the valley is compared with that of the Adige in Tyrol. Farther down, and as far as the neighbourhood of Ning-hia, a town built on the western banks of the Hoang-ho, at the eastern declivity of the Holang Shan (38° 32′ N. lat.), the valley has not been visited by Europeans. At this place the river runs in a wide valley which has been rendered fertile by numerous canals, which are fed by the waters of the river, and in which ice is extensively cultivated. There are also numerous plantations of fruit-trees. The soil contains much salt petre. The town of Ning-hia, the antient capital of Tangut, is of considerable extent, being fifteen li (equal to five miles) in cir

cuit. It has some very good manufactures of carpets and paper, and a considerable commerce with the nomadic tribes who wander about in the country west of the Holang Shan. Below the town of Ning-hia the valley of the Hoang-ho grows wider, as the range of the Holang-shan retires farther west, but its fertility decreases. About eighteen miles from Ning-hia the canals cease and no rice is cultivated. Other grain is still grown about 30 miles farther north, where the country gradually changes into a sandy, arid desert, interspersed with hills, swampy tracts, and pastures.

through which the road runs. The road leaves the valley of the Hoang-ho at the town of Lan-tcheou [CHINA, Vol. vii., p. 80], the capital of Kansi, and runs in a north-northwest direction over a stony and hilly country to the town of Liang-tcheou, a considerable place, of which however nothing is reported, except that the district in which it is situated is fertile, and contains a great number of villages. From Liang-tcheou the road runs north-west to Kan-tcheoufoo, a large and well-built town, which has many manufactures of woollen stuffs and felts, which articles are in great demand among the nomadic tribes of the Olöth Tshoros, who inhabit the contiguous part of the Gobi, and bring to the place their wool, horses, cattle, and sheep. It receives also large quantities of rhubarb from the Kilian Shan. From Kan-tcheou-foo the road continues in a north-west direction to So-tcheou, a large and well fortified town, with numerous bazars, well provided with provisions and manufactured articles. The town is divided into two sections, one of which is occupied by the Chinese, and the other by the foreign merchants from Bokhara and Turkistan. The latter is divided from the former by a separate wall, the gates of which are shut at night: in other respects foreigners do not experience any different treatment from natives. As So-tcheou is the last large place through which the caravans pass before they enter the desert between Tangut and Thian-shan-nahr: the commerce is very great, especially in provisions. About 50 or 60 miles west of So-tcheou is the most western gate of the Great Wall, called Kia-yu-kooan, or the gate of the You-stone (jasper), through which the caravans pass to enter the desert of Han-hai, which must be traversed in order to reach Hami in Thian-Shan-Nanlu. The last-mentioned town is 960 li, or 320 miles, from the gate of Kiayu-kooan, and that is the width of the Gobi at this place, which is considered the narrowest part of it.

The lateral valley of Si-ning-tcheou opens to the Hoangho from the west above the town of Lan-tcheou between the Kilian Shan and the most elevated portion of the Siue Shan. The valley is not extensive, but appears to be fertile it contains the town of Si-ning-tcheou, which is not quite as large as Ning-hia, but a much more commercial place, as the road which connects northern China with Hlassa in Tibet passes through it. This road leads from Si-ning-tcheou westward over a chain to the lake of Khookhoo-nor, which is of great but unknown extent. It is an alpine lake enclosed by high mountains, and has no outlet. The remainder of the road lies partly over numerous large mountain-masses, furrowed only by narrow glens and ravines, and partly over rocky and sandy tablelands, and the whole is described as a desert, in which only a small number of nomadic mountaineers are met with, and where the traveller for forty days' journey finds no other accommodation than the tents of the poor mountaineers. In spite of the difficulties, the road, as it appears, is much travelled, and the bazars of Si-ning-tcheou are well provided with provisions and articles of luxury. Even coffee and dates may be got there. This town is also the depôt of the Turkish rhubarb, which grows, as it appears, only on the more elevated parts of the Siue Shan and Kilian Shan, and is sent from Si-ning-tcheou to all parts of the world. Before the commerce between China and Siberia was established, this article was brought to Europe through Turkistan, Persia, and Turkey, and therefore is still called Turkey rhubarb, though at present it comes through Kiachta and Russia. When the Jesuits, who had been sent to these countries by the emperor Kang-hi, were at Si-a pass between high hills, through which a road leads ning-tcheou, they were astonished at seeing the quantity of rhubarb which, during the months of October and November, was daily brought from the adjacent mountains to the town.

The towns hitherto noticed lie along the great caravanroad, but farther west the Chinese geographers mention other places of importance. The largest, as it seems, is Ngan-si-foo, a town of the first rank, and the capital of the whole district. North-west of it, and on the border of the desert, is the town of Yu-men-kiang, which is built near northward to Hami, of which we have no information. South-west of Ngan-si-foot are the towns of Toong-hooangkiang, and Sha-tcheou. The last-mentioned place, whose name means Sandtown, seems to be the last inhabited place The northern part of Tangut, with the exception of the towards the west. It has not been visited by Europeans, valley of the Hoang-ho, is occupied by a wide desert plain, except by Marco Polo, who describes it as rather a large which constitutes a portion of the Gobi. [GOBI, vol. xi., place: he says that the inhabitants live on the produce of p. 286.] The steep declivities of the Kilian and Nan Shan their fields and orchards, and have little commerce. From however do not come close to the desert, but are separated his account, and that of a Chinese traveller, it is evident from it by a hilly tract from 30 to 50 miles wide, which that two roads run north-west and west from this place. contains some extensive tracts fit for cultivation, and Marco Polo reached it after traversing the desert of Lop, in which some large towns have been built, as the great by a thirty days' journey, having departed from the town commercial route which connects China with the coun- of Lop, which is on the banks of the lake of the same tries of Western Asia runs longitudinally through this name. The intermediate tract was mostly covered with hilly tract, and is confined to it by the extensive sandy sand, but in some places the soil consisted of bare and desert on the north, and the still less practicable mountain- broken rocks. A Chinese traveller departing from Shadesert which bounds it on the south. According to our tcheou, and taking the western route, seems to have trabest information, the ranges of the Kilian Shan, and espe-versed a still worse country, until he reached the town of cially of the Nan Shan, are covered with eternal snow, and one would imagine that they give origin to rivers which bring down a great volume of water, but that is not the case. The volume of water is very moderate: a part of it is consumed in irrigating the adjacent fields, and the remainder is absorbed by the sandy soil, as soon as it reaches the plain, after having left the hilly tract. This evidently shows that the watershed of the mountains must be at a very moderate distance from the Gobi. The surface of the hilly tract consists of an alternation of high lands and of depressions, running from the mountains northward to the border of the desert. The high lands are of considerable extent, their upper surface broken and rocky, and only occasionally covered with a thin layer of earth unfit for the growth of trees. In general the rocks are bare. The depressions between these high grounds are less extensive, but exhibit a considerable degree of fertility where they are irrigated. Even in those parts which are beyond the reach of irrigation, they are chiefly cultivated. To protect this hilly region, and the great commercial road which runs through it, against the nomadic tribes of the Gobi, the Chinese have continued the Great Wall along its northern border westward to 98° E. long., and along the wall are built the fortresses which protect the line and the towns

Khotan. [THIAN-SHAN-NANLU.]

That portion of the Gobi which lies north of the Great Wall contains many tracts which are covered with grass, and supply pasture to the Olöth Tshoros, but others have a sandy or stony soil, and are quite barren. In some places there are extensive swamps, especially where the rivers are lost, which descend from the Kilian Shan, among which the Etzina probably runs more than 200 miles. But the Han Hai, or that portion which lies between the gate of Kia-yu-kooan and Hami, is nearly uninhabited, as water is rarely met with, and the grassy tracts are still less frequent. The sand with which the surface is covered is very fine, and frequently raised into the air by strong winds.

Our information respecting the climate of Tangut is very scanty. The cold in winter is intense, and lasts for several months. The Jesuits found the Hoang-ho near 40° N. lat., at the end of November, covered with thick ice, so that the caravan was able to pass over it, though the river was more than 300 yards wide. At Ning-hia a heavy fall of snow was experienced in the middle of April. In summer the heat is great, but much less than in the low countries of China; the climate is considered as extremely healthy.

We are no better acquainted with the productions of