[Sidenote: Bill-discounting.]

Other incidents of the ordinary practice of banking are the discounting of bills, the keeping of deposit accounts, properly so called, and the making of advances to customers, counting either by way of definite loan or arranged overdraft. So far as the discounting of bills is concerned, there is little to differentiate the position of the banker from that of any ordinary bill-discounter. It has been contended, however, that the peculiar attribute of the banker's lien entitled him to hold funds of the customer against his liability on current discounted bills. This contention was ultimately disposed of by Bowen v. Foreign & Colonial Gas Company, 22 W.R. 740, where it was pointed out that the essential object of a customer's discounting bills with his banker was to feed the current account, and that a possible liability constituted no set-off against an existing debt. Whether a particular bill has been taken for discount or collection is a question of fact. As in the payment of bills, so in the collection of them, there is no statutory protection whatever for the banker; as against third parties he can only rely either on the customer's title or his own as a holder for value, if no forged endorsement intervene and he can establish a consideration.

[Sidenote: Deposit accounts.]

A deposit account, whether at call or on fixed notice, does not constitute any fiduciary relation between the depositor and the banker, but merely a debt due from the latter to the former. It has been suggested that cheques can be drawn against deposit account on call, and, though a banker might safely honour such a cheque, relying, if necessary, on his right of lien or set-off, there appears no legal right in the customer to enforce such payment. Deposit receipts given by bankers are exempt from stamp duty, even though they contain an undertaking with respect to payment of principal and interest. They are clearly not negotiable instruments, but it is difficult to deduce from the cases how far dealings with them may amount to an equitable assignment of the moneys they represent. Probably deliberate definite transfer, coupled with endorsement, would confer an effective title to such moneys. Where, as is not uncommon, the form of deposit note includes a cheque, the banker could not refuse to pay were the cheque presented and any superadded formalities complied with.

[Sidenote: Overdrafts and advances.]

There is no obligation on a banker to permit his customer to overdraw, apart from agreement express or implied from course of business. Drawing a cheque or accepting a bill payable at the banker's which there are not funds meet is an implied request for an overdraft, which the banker may or may not comply with. Interest is clearly chargeable on overdrafts whether stipulated for or not. There is no direct authority establishing this right in the banker, and interest is not usually recoverable on mere debts, but the charge is justifiable on the ground of the universal custom of bankers, if not otherwise. The charging of compound interest or interest with periodical rests has been supported where such system of keeping the accounts has been brought to the notice of the customer by means of the pass-book, and not objected to by him, but in the present attitude of the courts towards the pass-book some further recognition would seem necessary. Such system of charging interest, even when fully recognized, only prevails so long as the relation of banker and customer, on which it is founded, continues in force; the taking a mortgage for the existing debt would put an end to it.

[Sidenote: Lien.]

The main point in which advances made by bankers differ from those made by other people is the exceptional right possessed by bankers of securing repayment by means of the banker's lien. The banker's lien is part of the law merchant and entitles him, in the absence of agreement express or implied to the contrary, to retain and apply, in discharge of the customer's liability to him, any securities of the customer coming into his possession in his capacity as banker. It includes bills and cheques paid in for collection (Currie v. Misa, 1 A.C. 564). Either by virtue of it, or his right of set-off, the banker can retain moneys paid in by or received for the credit of the customer, against the customer's debt to him. Goods deposited for safe custody or moneys paid in to meet particular bills are exempt from the lien, the purpose for which they come to the banker's hands being inconsistent with the assertion of the lien. The existence of the banker's lien entitles him to sue all parties to bills or cheques by virtue of sec. 27, subs. 3 of the Bills of Exchange Act, and to the extent of his advances his title is independent of that of the previous holder. Moreover, the banker's lien, though so termed, is really in effect an implied pledge, and confers the rights of realization on default pertaining to that class of bailment. But with regard to the exercise of his lien, as in many other phases of his relation to his customer, the banker's strict rights may be curtailed or circumscribed by limitations arising out of course of business. The principle, based either on general equity or estoppel and independent of definite agreement or consideration, requires that when dealings between banker and customer have for a reasonable space of time proceeded on a recognized footing, the banker shall not suddenly break away from such established order of things and assert his strict legal rights to the detriment of the customer. By the operation of this rule, the banker may be precluded from asserting his lien in particular cases, as for instance for an overdraft on one account against another which had habitually been kept and operated on separately. It equally prevents the dishonouring of cheques in circumstances in which they have hitherto been paid independent of the actual available balance.

Restrictions arising from course of business can of course be put an end to by the banker, but only on reasonable notice to the customer and by providing for outstanding liabilities undertaken by the latter in reliance on the continuance of the pre-existing state of affairs (see Buckingham v. London & Midland Bank, 12 Times L.R. 70). As against this, the banker can, in some cases, fortify his position by appeal to the custom of bankers. The validity of such custom, provided it be general and reasonable, has frequently been recognized by the courts. Any person entering on business relations with a banker must be taken to contemplate the existence of such custom and implicitly agree that business shall be conducted in accordance therewith. Practical difficulty has been suggested with regard to proof of any such custom not already recognized in law, as to how far it can be established by the evidence of one party, the bankers, unsupported by that of members of the outside public, in most cases impossible to obtain. It is conceived, however, that on the analogy of local custom and the Stock Exchange rules, such outside evidence could be dispensed with, and this is the line apparently indicated with relation to the pass-book by the court of appeal in Vagliano's case (23 Q.B.D. at p. 245). The unquestionable right of the banker to summarily debit his customer's account with a returned cheque, even when unindorsed by the customer and taken by the banker in circumstances constituting him a transferee of the instrument, is probably referable to a custom of this nature. So is the common practice of bankers to refuse payment of a so-called "stale" cheque, that is, one presented an unreasonable time after its ostensible date; although the fact that some banks treat a cheque as stale after six months, others not till after twelve, might be held to militate against the validity of such custom, and lapse of time is not included by the Bills of Exchange Act among the matters working revocation of the banker's duty, and authority to pay his customer's cheque. Indirectly, this particular custom obtains some support from sec. 74 (2) of the Bills of Exchange Act, although the object of that section is different.

That section does, however, import the custom of bankers into the reckoning of a reasonable time for the presentation of a cheque, and with other sections clears up any doubts which might have arisen on the common law as to the right of the holder of a cheque, whether crossed or not, to employ his banker for its collection, without imperilling his rights against prior parties in case of dishonour. On dishonour of a cheque paid in for [v.03 p.0353] collection, the banker is bound to give notice of dishonour. Being in the position of an agent, he may either give notice to his principal, the customer, or to the parties liable on the bill. The usual practice of bankers has always been to return the cheque to the customer, and sec. 49, subs. 6 of the Bills of Exchange Act is stated to have been passed to validate this custom. Inasmuch as it only provides for the return of the dishonoured bill or cheque to the drawer or an endorser it appears to miss the case of a cheque to bearer or become payable to bearer by blank endorsement prior to the customer's.

Where a bank or a banker takes a mortgage, legal or equitable, or a guarantee as cover for advances or overdraft, there is nothing necessarily differentiating the position from that of any other mortgagee or guaranteed party. It has, however, fallen to banks to evoke some leading decisions with respect to the former class of security. In London Joint Stock Bank v. Simmons ([1892], A.C. 201) the House of Lords, professedly explaining their previous decision in Sheffield v. London Joint Stock Bank, 13 A.C. 333, determined that negotiable securities, commercial or otherwise, may safely be taken in pledge for advances, though the person tendering them is, from his known position, likely to be holding them merely as agent for other persons, so long as they are taken honestly and there is nothing tangible, outside the man's position, to arouse suspicion. So again in Lloyd's Bank v. Cooke [1907], 1 K.B. 794, the bank vindicated the important principle that the common law of estoppel still obtains with regard to bills, notes and cheques, save where distinctly annulled or abrogated by the Bills of Exchange Act, and that therefore a man putting inchoate negotiable instruments into the hands of an agent for the purpose of his raising money thereon is responsible to any one taking them bona fide and for value, although the agent may have fraudulently exceeded and abused his authority and the case does not fall within the provisions of the Bills of Exchange Act.

[Sidenote: Guarantees.]

With regard to guarantees, the main incidents peculiarly affecting bankers are the following. The existence of a guarantee does not oblige the banker to any particular system of keeping the account. So long as it is not unfairly manipulated to the detriment of the guarantor, there is no obligation to put moneys paid in, without appropriation, to the guaranteed rather than to the unguaranteed account, and on the termination of a guarantee, the banker may close the account, leaving it to be covered by the guarantee, and open a new one with the customer, to which he may devote payments in, not otherwise appropriated. Where by its nature or terms a continuing guarantee is revocable either summarily or on specified notice, difficult questions may arise on such revocation as to the banker's duty and obligations towards the customer, who has probably incurred liabilities on the strength of the credit afforded by the guarantee. Although the existence of a guarantee does not bind the banker to advance up to the prescribed limit, he could not well, on revocation, immediately shut off all facilities from the customer without notice, while subsequent purely voluntary advances might not be covered by the guarantee. These contingencies should therefore be fully provided for by the guarantee, particularly the crucial period of the pendency of notice.

AUTHORITIES.—The Institute of Bankers (London), Questions on Banking Practice (6th ed., 1909); J. Douglas Walker, A Treatise on Banking Law (2nd ed., 1885); Chalmers, Bills of Exchange (7th ed., 1909); Sir J. R. Paget, The Law of Banking (2nd ed., 1908); H. Hart, The Law of Banking (2nd ed., 1906).

(J. R. P.)

[1] A translation of the act of the 3rd of May 1619 may be found in the appendix to the Quarterly Journal of Economics (Boston, U.S.A.) for April 1892. These documents present a distinct picture of banking in its true sense.

[2] The clearest account of its early days is found in Thorold Rogers' History of the First Nine Years of the Bank of England.

[3] The date 1876 is taken as being that when the Imperial Bank of Germany came into full operation.

[4] "The Grasshopper" in Lombard Street, by John Biddulph Masters (1892).

[5] See Vortraege und Aufsaetze hauptsaechlich aus dem Handels- und Wechselrecht, von Dr R. Koch, pp. 163-164.

[6] The imperial treasury is bound to pay the state notes in cash at any time when this is required, but an independent fund of cash set apart for this purpose does not exist. See Handwoerterbuch der Staatswissenschaften, vol. v. art. "Papiergeld," p. 97 (Jena, 1893; ed. J. Conrad, L. Elster, W. Lexis and E. Loening).

BANKSIA, an Australian genus of shrubs and trees (natural order Proteaceae), with leathery leaves often deeply cut and handsome dense spikes of flowers. It is named after Sir Joseph Banks (q.v.). The plants are grown in England for their handsome foliage as evergreen greenhouse shrubs.

BANKURA, a town and district of British India, within the Burdwan division of Bengal. The town has a population of 20,737. The district has an area of 2621 sq. m., and in 1901 its population was 1,116,411, showing an increase of 4% in the decade. It is bounded on the N. and E. by Burdwan district; on the S. by Midnapur district; and on the W. by Manbhum district. Bankura forms a connecting link between the delta of the Ganges on the E. and the mountainous highlands of Chota Nagpur on the W. Along its eastern boundary adjoining Burdwan district the country is flat and alluvial, presenting the appearance of the ordinary paddy lands of Bengal. Going N. and W., however, the surface gradually rises into long undulating tracts; rice lands and swamps give way to a region of low thorny jungle or forest trees; the hamlets become smaller and more scattered, and nearly disappear altogether in the wild forests along the western boundary. Large quantities of lac and tussur silk are gathered in the hilly tract. The stone quarries and minerals are little worked. There are indigo factories and two coal-mines. Both cotton and silk are woven, and plates, &c., are carved from soap-stone. The old capital of the country was at Bishnupur, which is still the chief centre of local industries. The north-east part of the district is skirted by the East Indian railway beyond the river Damodar. The Midnapur-Jherria line of the Bengal-Nagpur railway passes through the district, and there is a line from Howrah to Bankura. The climate of Bankura is generally healthy, the cold season being bracing, the air wholesome and dry, and fogs of rare occurrence. The district is exposed to drought and also to destructive floods. It suffered in the famines of 1866, 1874-1875 and 1896-1897. The temperature in the hot season is very oppressive and relaxing. The Bishnupur raj was one of the largest estates in Bengal in the end of the 18th century, but it was sold for arrears of revenue shortly after the conclusion of the permanent settlement in 1793.

BANN, the principal river in the north of Ireland. Rising in the Mourne mountains in the south of the Co. Down it runs N.W. until it enters Lough Neagh (q.v.), which it drains N.N.W. to an estuary at Coleraine, forming Lough Beg immediately below the larger lough. The length of its valley (excluding the lesser windings of the river) is about 90 m. The total drainage area, including the other important feeders of Lough Neagh, is about 2300 sq. m., extending westward to the confines of the Co. Fermanagh, and including parts of the Cos. Down and Antrim, Armagh and Monaghan, Tyrone and Londonderry. The river has valuable salmon fisheries, but is not of much importance for navigation. Above Lough Neagh it is known as the Upper Bann and below as the Lower Bann.

BANNATYNE, GEORGE (1545-?1608), collector of Scottish poems, was a native of Newtyle, Forfarshire. He became an Edinburgh merchant and was admitted a burgess in 1587. Some years earlier, in 1568, when the "pest" raged in the capital, he retired to his native county and amused himself by writing out copies of poems by 15th and early 16th century Scots poets. His work extended to eight hundred folio pages, divided into five parts. The MS. descended to his only daughter Janet, and later to her husband's family, the Foulises of Woodhall and Ravelston, near Edinburgh. From them it passed to the Advocates' library, where it is still preserved. This MS., known as the "Bannatyne Manuscript," constitutes with the "Asloan" and "Maitland Folio" MSS. the chief repository of Middle Scots poetry, especially for the texts of the greater poets Henryson, Dunbar, Lyndsay and Alexander Scott. Portions of it were reprinted (with modifications) by Allan Ramsay in his Ever Green (1724), and later, and more correctly, by Lord Hailes in his Ancient Scottish Poems (1770). The entire text was issued by the Hunterian Club (1873-1902) in a handsome and generally accurate form. The name of Bannatyne was honoured in 1823 by the foundation in Edinburgh of the Bannatyne Club, devoted to the publication of historical and literary material from Scottish sources. The thirty-third issue of the club (1829) was Memorials of George Bannatyne (1545-1608), with a memoir by Sir Walter Scott and an account of the MS. by David Laing.

See also Gregory Smith, Specimens of Middle Scots (1902).

BANNERET (Fr. banneret, from banniere, banner, elliptical for seigneur or chevalier banneret, Med. Lat. banneretus), in feudalism, the name given to those nobles who had the right to lead their vassals to battle under their own banner. Ultimately bannerets obtained a place in the feudal hierarchy between [v.03 p.0354] barons and knights bachelors, which has given rise to the idea that they are the origin of King James I.'s order of baronets. Selden, indeed, points out that "the old stories" often have baronetti for bannereti, and he points out that in France the title had become hereditary; but he himself is careful to say (p. 680) that banneret "hath no relation to this later title." The title of knight banneret, with the right to display the private banner, came to be granted for distinguished service in the field. "No knight banneret," says Selden, of the English custom, "can be created but in the field, and that, when either the king is present, or at least his royal standard is displayed. But the creation is almost the self-same with that in the old French ceremonies by the solemn delivery of a banner charged with the arms of him that is to be created, and the cutting of the end of the pennon or streamer to make it a square or into the shape of a banner in case that he which is to be created had in the field his arms on a streamer before the creation." The creation of bannerets is traceable, according to Selden, to the time of Edward I. "Under these bannerets," he adds, "divers knights bachelors and esquires usually served; and according to the number of them, the bannerets received wages." The last authentic instance of the creation of a knight banneret was that of John Smith, created banneret at the battle of Edgehill by Charles I. for rescuing the royal standard from the enemy.

See Selden, Titles of Honor (3rd ed., London, 1672), p. 656; Du Cange, Glossarium (Niort, 1883), s.v. "Bannereti."

BANNERS, FEAST OF (Jap. Nobori-no-Sekku), a Japanese festival in honour of male children held on the 5th of May. Every householder who has sons fastens a bamboo pole over his door and hangs from it gaily-coloured paper fishes, one for each of his boys. These fishes are made to represent carp, which are in Japanese folklore symbolical of health and longevity. The day is recognized as a national holiday.

For banners in general see FLAG.

BANNISTER, CHARLES (1738-1804), English actor and singer, was born in Gloucestershire, and after some amateur and provincial experience made his first London appearance in 1762 as Will in The Orators at the Haymarket. Gifted with a fine bass voice, Bannister acquired a reputation as a singer at Ranelagh and elsewhere, as well as an actor, and was received with such favour that Garrick engaged him for Drury Lane. He died on the 26th of October 1804.

His son JOHN BANNISTER (1760-1836), born at Deptford on the 12th of May 1760, first studied to be a painter, but soon took to the stage. His first formal appearance was at the Haymarket in 1778 as Dick in The Apprentice. The same year at Drury Lane he played in James Miller's version of Voltaire's Mahomet the part of Zaphna, which he had studied under Garrick. The Palmira of the cast was Mrs Robinson ("Perdita"). Bannister was the best low comedian of his day. As manager of Drury Lane (1802) he was no less successful. He retired in 1815 and died on the 7th of November 1836. He never gave up his taste for painting, and Gainsborough, Morland and Rowlandson were among his friends.

See Adolphus's Memoirs of John Bannister (2 vols., 1838).

BANNOCK (adapted from the Gaelic, and apparently connected with Lat. panis, bread), the term used in Scotland and the north of England for a large, flattish, round sort of bun or cake, usually made of barley-meal, but also of wheat, and sometimes with currants.

BANNOCK, the name of a county in the south-east of the state of Idaho, U.S.A., and of a river in the same state, which runs northward in Oneida county into the Snake or Lewis river. It is taken from that of the Bannock Indians (see BANATE), a corruption of the native Panaiti.

BANNOCKBURN, a town of Stirlingshire, Scotland. Pop. (1901) 2444. It is situated on the "burn" from which its name is derived, the Bannock (Gaelic, ban oc, "white, shining stream"), a right-hand affluent of the Forth, which was once a considerable river. The town lies 2-1/4 m. S.S.E. of Stirling by the Caledonian railway, and now has thriving manufactures of woollens (chiefly tweeds, carpets and tartans) and leather, though at the beginning of the 19th century it was only a village. The Bore Stone, in which Bruce planted his standard before the battle in which he defeated Edward II. in 1314 (see below), is preserved by an iron grating. A mile to the west is the Gillies' Hill, now finely wooded, over which the Scots' camp-followers appeared to complete the discomfiture of the English, to which event it owes its name. Bannockburn House was Prince Charles Edward's headquarters in January 1746 before the fight at Falkirk.

The famous battle of Bannockburn (24th June 1314) was fought for the relief of Stirling Castle, which was besieged by the Scottish forces under Robert Bruce. The English governor of Stirling had promised that, if he were not relieved by that date, he would surrender the castle, and Edward II. hastily collected an army in the northern and midland counties of England. Bruce made no attempt to defend the border, and selected his defensive position on the Bannock Burn, 2-1/2 m. S. of Stirling. His front was covered by the marshy bed of the stream, his left flank by its northerly bend towards the Forth, his right by a group of woods, behind which, until the English army appeared, the Scots concealed themselves. Two corps were left in the open in observation, one at St Ninian's to watch the lower course of the burn, one to guard the point at which the Falkirk-Stirling road crosses the burn. On the 23rd the van of the army of Edward, which numbered about 60,000 against the 40,000 of the Scots, appeared to the south of the burn and at once despatched two bodies of men towards Stirling, the first by the direct road, the other over the lower Bannock Burn near its junction with the Forth. The former was met by the Scottish outpost on the road, and here occurred the famous single combat in which Robert Bruce, though not fully armed for battle, killed Sir Henry Bohun. The English corps which took the other route was met and after a severe struggle defeated by the second Scottish outpost near St Ninian's. The English army assembled for battle on the following day. Early on St John's day the Scottish army took up its assigned positions. Three corps of pikemen in solid masses formed the first line, which was kept out of sight behind the crest until the enemy advanced in earnest. A line of "pottes" (military pits) had been previously dug to give additional protection to the front, which extended for about one mile from wing to wing. The reserve under Bruce consisted of a corps of pikemen and a squadron of 500 chosen men-at-arms under Sir Robert Keith, the marischal of Scotland. The line of the defenders was unusually dense; Edward, in forming up on an equal front with greatly superior numbers, found his army almost hopelessly cramped. The attacking army was formed in an unwieldy mass of ten "battles," each consisting of horse and foot, and the whole formed in three lines each of three "battles," with the tenth "battle" as a reserve in rear. In this order the English moved down into the valley for a direct attack, the cavalry of each "battle" in first line, the foot in second. Ignoring the lesson of Falkirk (q.v.), the mounted men rode through the morass and up the slope, which was now crowned by the three great masses of the Scottish pikemen. The attack of the English failed to make any gap in the line of defence, many knights and men-at-arms were injured by falling into the pits, and the battle became a melee, the Scots, with better fortune than at Falkirk and Flodden, presenting always an impenetrable hedge of spears, the English, too stubborn to draw off, constantly trying in vain to break it down. So great was the press that the "battles" of the second line which followed the first were unable to reach the front and stood on the slope, powerless to take part in the battle on the crest. The advance of the third English line only made matters worse, and the sole attempt to deploy the archers was crushed with great slaughter by the charge of Keith's mounted men. Bruce threw his infantry reserve into the battle, the arrows of the English archers wounded the men-at-arms of their own side, and the remnants of the leading line were tired and disheartened when the final impetus to their rout was given by the historic charge of the "gillies," some thousands of Scottish camp-followers who suddenly emerged from the woods, blowing horns, waving such weapons as they possessed, and holding aloft [v.03 p.0355] improvised banners. Their cries of "slay, slay!" seemed to the wearied English to betoken the advance of a great reserve, and in a few minutes the whole English army broke and fled in disorder down the slope. Many perished in the burn, and the demoralized fugitives were hunted by the peasantry until they re-crossed the English border. One earl, forty-two barons and bannerets, two hundred knights, seven hundred esquires and probably 10,000 foot were killed in the battle and the pursuit. One earl, twenty-two barons and bannerets and sixty-eight knights fell into the hands of the victors, whose total loss of 4000 men included, it is said, only two knights.

See J. E. Shearer, Fact and Fiction in the Story of Bannockburn (1909).

BANNS OF MARRIAGE (formerly bannes, from A.S. gebann, proclamation, Fr. ban, Med. Lat. bannum), the public legal notice of an impending marriage. The church in earliest days was forewarned of marriages (Tertullian, Ad Uxorem, De Pudicitia, c. 4). The first canonical enactment on the subject in the English church is that contained in the 11th canon of the synod of Westminster in London (A.D. 1200), which orders that "no marriage shall be contracted without banns thrice published in the church, unless by special authority of the bishop." It is, however, believed that the practice was in France as old as the 9th century, and certainly Odo, bishop of Paris, ordered it in 1176. Some have thought that the custom originated in the ancient rule that all "good knights and true," who elected to take part in the tournaments, should hang up their shields in the nearest church for some weeks before the opening of the lists, so that, if any "impediment" existed, they might be "warned off." By the Lateran Council of 1215 the publication of banns was made compulsory on all Christendom. In early times it was usual for the priest to betroth the pair formally in the name of the Blessed Trinity; and sometimes the banns were published at vespers, sometimes during mass. In the United Kingdom, under the canon law and by statute, banns are the normal preliminary to marriage; but a marriage may also be solemnized without the publication of banns, by obtaining a licence or a registrar's certificate. In America there is no statutory requirement; and the practice of banns (though general in the colonial period) is practically confined to the Roman Catholics.

BANNU, a town and district of British India, in the Derajat division of the North-West Frontier Province. The town (also called Edwardesabad and Dhulipnagar) lies in the north-west corner of the district, in the valley of the Kurram river. Pop. (1901) 14,300. It forms the base for all punitive expeditions to the Tochi Valley and Waziri frontier.

The district of Bannu, which only consists of the Bannu and Marwat tahsils since the constitution of the North-West Frontier Province in 1901, contains an area of 1680 sq. m. lying north of the Indus. The cis-Indus portions of Bannu and Dera Ismail Khan now comprises the new Punjab district of Mianwali. In addition to the Indus the other streams flowing through the district are the Kurram (which falls into the Indus) and its tributary the Gambila. The valley of Bannu proper, stretching to the foot of the frontier hills, forms an irregular oval, measuring 60 m. from north to south and about 40 m. from east to west. In 1901 the population was 231,485, of whom the great majority were Mahommedans. The principal tribes inhabiting the district are: (1) Waziri Pathans, recent immigrants from the hills, for the most part peaceable and good cultivators; (2) Marwats, a Pathan race, inhabiting the lower and more sandy portions of the Bannu valley; (3) Bannuchis, a mongrel Afghan tribe of bad physique and mean vices. The inhabitants of this district have always been very independent and stubbornly resisted the Afghan and Sikh predecessors of the British. After the annexation of the Punjab the valley was administered by Herbert Edwardes so thoroughly that it became a source of strength instead of weakness during the Mutiny. The inhabitants of the valley itself are now peaceful, but it is always subject to incursion from the Waziri tribes in the Tochi valley and the neighbouring hills. Salt is quarried on government account at Kalabagh and alum is largely obtained in the same neighbourhood. The chief export is wheat. A military road leads from Bannu town towards Dera Ismail Khan. The Indus, which is nowhere bridged within the district, is navigable for native boats throughout its course of 76 m. The chief frontier tribes on the border are the Waziris, Battannis and Dawaris. All these are described under their separate names.

BANSDA, a native state in the south Gujarat division of Bombay, India, belonging to the Surat agency. Area, 215 sq. m. Pop. (1901) 40,382, showing a decrease of 2% in the decade; estimated revenue L19,508. Its chief is a rajput. About half the total area of the state is cultivable, but the bulk is forested.

BANSHEE (Irish bean sidhe; Gaelic ban sith, "woman of the fairies"), a supernatural being in Irish and general Celtic folklore, whose mournful screaming, or "keening," at night is held to foretell the death of some member of the household visited. In Ireland legends of the banshee belong more particularly to certain families in whose records periodic visits from the spirit are chronicled. A like ghostly informer figures in Brittany folklore. The Irish banshee is held to be the distinction only of families of pure Milesian descent. The Welsh have the banshee under the name gwrach y Rhibyn (witch of Rhibyn). Sir Walter Scott mentions a belief in the banshee as existing in the highlands of Scotland (Demonology and Witchcraft, p. 351). A Welsh death-portent often confused with the gwrach y Rhibyn and banshee is the cyhyraeth, the groaning spirit.

See W. Wirt Sikes, British Goblins (1880).

BANSWARA (literally "the forest country"), a rajput feudatory state in Rajputana, India. It borders on Gujarat and is bounded on the N. by the native states of Dungarpur and Udaipur or Mewar; on the N.E. and E. by Partabgarh; on the S. by the dominions of Holkar and the state of Jabua and on the W. by the state of Rewa Kantha. Banswara state is about 45 m. in length from N. to S., and 33 m. in breadth from E. to W., and has an area of 1946 sq. m. The population in 1901 was 165,350. The Mahi is the only river in the state and great scarcity of water occurs in the dry season. The Banswara chief belongs to the family of Udaipur. During the vigour of the Delhi empire Banswara formed one of its dependencies; on its decline the state passed under the Mahrattas. Wearied out by their oppressions, its chief in 1812 petitioned for English protection, on the condition of his state becoming tributary on the expulsion of the Mahrattas. The treaty of 1818 gave effect to this arrangement, Britain guaranteeing the prince against external enemies and refractory chiefs; he, on his part, pledging himself to be guided by her representative in the administration of his state. The chief is assisted in the administration by a hamdar or minister. The estimated gross revenue is L17,000 and the tribute L2500. The custom of suttee, or widow-burning, has long been abolished in the state, but the people retain all their superstitions regarding witches and sorcery; and as late as 1870, a Bhil woman, about eighty years old, was swung to death at Kushalgarh on an accusation of witchcraft. The perpetrators of the crime were sentenced to five years' rigorous imprisonment, but they had the sympathy of the people on their side. The chief town is Banswara, situated about 8 m. W. of the Mahi river, surrounded by an old disused rampart and adorned by various Hindu temples, with the battlements of the chief's palace overlooking it. Its population in 1901 was 7038. The petty state of Kushalgarh is feudatory to Banswara.

BANTAM, the westernmost residency of the island of Java, Dutch East Indies, bounded W. by the Strait of Sunda, N. by the Java sea, E. by the residencies of Batavia and Preanger, and S. by the Indian Ocean. It also includes Princes Island and Dwars-in-den-weg ("right-in-the-way") Island in Sunda Strait, as well as several smaller islands along the coasts. Bantam had a population in 1897 of 709,339, including 302 Europeans, 1959 Chinese and 89 Arabs and other Asiatic foreigners. The natives are Sundanese, except in the northern or Serang division, where they are Javanese. The coast is low-lying and frequently marshy. The northern portion of the residency constitutes the most fertile portion, is generally flat with a hilly group in the middle, where the two inactive volcanoes, Karang and Pulosari, [v.03 p.0356] are found, while the north-western corner is occupied by the isolated Gede Mountain. The southern portion is covered by the Kendang (Malay for "range") Mountains extending into the Preanger. The rivers are only navigable at their mouths. Various geysers and cold and warm sulphur springs are found in the centre of the residency, and on a ridge of the Karang Mountain is the large crater-lake Dano, a great part of which was drained by the government in 1835 for rice cultivation. Pulse (kachang), rice and coffee are the principal products of cultivation; but in the days of government culture sugar, indigo and especially pepper were also largely grown. The former considerable fishing and coasting trade was ruined by the eruption of Krakatoa in 1883, a large stretch of coast line and the seaport towns of Charingin and Anjer being destroyed by the inundation. The prosperity of the residency was further affected by a cattle plague in 1879, followed by a fever epidemic which carried off 50,000 people, and except in the rice season there is a considerable emigration of natives. Bantam contains five native regencies or territorial divisions, namely, Serang, Anjer, Pandeglang, Charingin, Lebak. The principal towns are Serang, the capital of the residency, Chilegon, Pandeglang, Menes and Rangkas Betug. The chief town, Serang, is situated 2 1/2 m. from Bantam Bay on the high road from Batavia. The port of Serang is Karangantu, on Bantam Bay, and close by is the old ruined town of Bantam, once the capital of the kingdom of Bantam, and before the foundation of Batavia the principal commercial port of the Dutch East India Company. The ruins include the remains of the former pepper warehouses, the old factory, called Fort Speelwijk, belonging to the company, the fortified palace of the former sultans and a well-preserved mosque thought to have been built by the third Mahommedan ruler of Bantam about 1562-1576, and containing the tombs of various princes of Bantam. Before the Dutch conquest Bantam was a powerful Mahommedan state, whose sovereign extended his conquests in the neighbouring islands of Borneo and Sumatra. In 1595 the Dutch expelled the Portuguese and formed their first settlement. A British factory was established in 1603 and continued to exist till the staff was expelled in 1682. In 1683 the Dutch reduced the sultan to vassalage, built the fort of Speelwijk and monopolized the port, which had previously been free to all comers; and for more than a century afterwards Bantam was one of the most important seats of commerce in the East Indies. In 1811 after Batavia had surrendered to the British, Bantam soon followed; but it was restored to the Dutch in 1814. Two years later, however, they removed their chief settlement to the more elevated station of Serang, or Ceram, 7 m. inland, and in 1817 the ruin of Bantam was hastened by a fire.

For "Bantam" fowls see POULTRY.

BANTIN, or BANTING, the native name of the wild ox of Java, known to the Malays as sapi-utan, and in zoology as Bos (Bibos) sondaicus. The white patch on the rump distinguishes the bantin from its ally the gaur (q.v.). Bulls of the typical bantin of Java and Borneo are, when fully adult, completely black except for the white rump and legs, but the cows and young are rufous. In Burma the species is represented by the tsaine, or h'saine, in which the colour of the adult bulls is rufous fawn. Tame bantin are bred in Bali, near Java, and exported to Singapore. (See BOVIDAE.)

BANTRY, a seaport, market-town and seaside resort of Co. Cork, Ireland, in the west parliamentary division, 58 m. S.W. of Cork by the Cork, Bandon & South Coast railway, on the bay of the same name. Pop. (1901) 3109. It is an important centre both for sea fisheries and for sport with the rod. It is the terminus of the railway, and a coaching station on the famous "Prince of Wales" route (named after King Edward VII.) from Cork to Glengarriff and Killarney. The bay, with excellent anchorage, is a picturesque inlet some 22m. long by 3 to 6 broad, with 12 to 32 fathoms of water. It is one of the headquarter stations of the Channel Squadron, which uses the harbour at Castletown Bearhaven on the northern shore, behind Bear Island, near the mouth of the bay. It was the scene of attempts by the French to invade Ireland in 1689 and 1796, and troops of William of Orange were landed here in 1697. There are several islands, the principal of which are Bear Island and Whiddy, off the town. Ruins of the so-called "fish palaces" testify to the failure of the pilchard fishery in the 18th century.

BANTU LANGUAGES. The greater part of Africa south of the equator possesses but one linguistic family so far as its native inhabitants are concerned. This clearly-marked division of human speech has been entitled the Bantu, a name invented by Dr W. H. I. Bleek, and it is, on the whole, the fittest general term with which to designate the most remarkable group of African languages.[1]

It must not be supposed for a moment that all the people who speak Bantu languages belong necessarily to a special and definite type of negro. On the contrary, though there is a certain physical resemblance among those tribes who speak clearly-marked Bantu dialects (the Babangi of the upper Congo, the people of the Great Lakes, the Ova-herero, the Ba-tonga, Zulu-Kaffirs, Awemba and some of the East Coast tribes), there is nevertheless a great diversity in outward appearance, shape of head and other physical characteristics, among the negroes who inhabit Bantu Africa. Some tribes speaking Bantu languages are dwarfs or dwarfish, and belong to the group of Forest Pygmies. Others betray relationship to the Hottentots; others again cannot be distinguished from the most exaggerated types of the black West African negro. Yet others again, especially on the north, are of Gala (Galla) or Nilotic origin. But the general deduction to be drawn from a study of the Bantu languages, as they exist at the present day, is that at some period not more than 3000 years ago a powerful tribe of negroes speaking the Bantu mother-language, allied physically to the negroes of the south-western Nile and southern Lake Chad basins (yet impregnated with the Caucasian Hamite), pushed themselves forcibly from the very heart of Africa (the region between the watersheds of the Shari, Congo and western Nile) into the southern half of the continent, which at that time was probably sparsely populated except in the north-west, east and south. The Congo basin and the south-western watershed of the Nile at the time of the Bantu invasion would have been occupied on the Atlantic seaboard by West Coast negroes, and in the centre by negroes of a low type and by Forest Pygmies; the eastern coasts of Victoria Nyanza and the East African coast region down to opposite Zanzibar probably had a population partly Nilotic-negro and partly Hottentot-Bushman. From Lakes Tanganyika and Nyasa south-westwards to the Cape of Good Hope the population was Forest-negro, Nilotic-negro, Hottentot and Bushman. Over nearly all this area the Bantu swept; and they assimilated or absorbed the vast majority of the preceding populations, of which, physically or linguistically, the only survivors are the scattered tribes of pygmies in the forests of south-west Nile land, Congo basin and Gabun, the central Sudanese of the N.E. Congo, a few patches of quasi-Hottentot, Hamitic and Nilotic peoples between Victoria Nyanza and the Zanzibar coast, and the Bushmen and Hottentots of south-west Africa. The first area of decided concentration on the part of the Bantu was very probably Uganda and the shores of Tanganyika. The main line of advance south-west trended rather to the east coast of Africa than to the west, but bifurcated at the south end of Lake Tanganyika, one great branch passing west between that lake and Nyasa, and the other southwards. Finally, when the Bantu had reached the [v.03 p.0357] south-west corner of Africa, their farther advance was checked by two causes: first, the concentration in a healthy, cattle-rearing part of Africa of the Hottentots (themselves only a superior type of Bushman, but able to offer a much sturdier resistance to the big black Bantu negroes than the crafty but feeble Bushmen), and secondly, the arrival on the scene of the Dutch and British, but for whose final intervention the whole of southern Africa would have been rapidly Bantuized, as far as the imposition of language was concerned.

The theory thus set forth of the origin and progress of the Bantu and the approximate date at which their great southern exodus commenced, is to some extent attributable to the present writer only, and has been traversed at different times by other writers on the same subject. In the nearly total absence of any historical records, the only means of building up Bantu history lies in linguistic research, in the study of existing dialects, of their relative degree of purity, of their connexion one with the other and of the most widely-spread roots common to the majority of the Bantu languages. The present writer, relying on linguistic evidence, fixed the approximate date at which the Bantu negroes left their primal home in the very heart of Africa at not much more than 2000 years ago; and the reason adduced was worth some consideration. It lay in the root common to a large proportion of the Bantu languages expressing the domestic fowl—kuku (nkuku, ngoko, nsusu, nguku, nku). Now the domestic fowl reached Africa first through Egypt, at the time of the Persian occupation—not before 500 to 400 B.C. It would take at that time at least a couple of hundred years before—from people to people and tribe to tribe up the Nile valley—the fowl, as a domestic bird, reached the equatorial regions of Africa. The Muscovy duck, introduced by the Portuguese from Brazil at the beginning of the 17th century, is spreading itself over Negro Africa at just about the same rate. Yet the Bantu people must have had the domestic fowl well established amongst themselves before they left their original home, because throughout Bantu Africa (with rare exceptions and those not among the purest Bantu tribes) the root expressing the domestic fowl recurs to the one vocable of kuku.[2] Curiously enough this root kuku resembles to a marked degree several of the Persian words for "fowl," and is no doubt remotely derived from the cry of the bird. Among those Negro races which do not speak Bantu languages, though they may be living in the closest proximity to the Bantu, the name for fowl is quite different.[3] The fowl was only introduced into Madagascar, as far as researches go, by the Arabs during the historical period, and is not known by any name similar to the root kuku. Moreover, even if the fowl had been (and there is no record of this fact) introduced from Madagascar on to the east coast of Africa, it would be indeed strange if it carried with it to Cameroon, to the White Nile and to Lake Ngami one and the same name. It may, however, be argued that such a thing is possible, that the introduction of the fowl south of the equator need not be in any way coincident with the Bantu invasion, as its name in North Central Africa may have followed it everywhere among the Bantu peoples. But all other cases of introduced plants or animals do not support this idea in the least. The Muscovy duck, for instance, is pretty well distributed throughout Bantu Africa, but it has no common widely-spread name. Even tobacco (though the root "taba" turns up unexpectedly in remote parts of Africa) assumes totally different designations in different Bantu tribes. The Bantu, moreover, remained faithful to a great number of roots like "fowl," which referred to animals, plants, implements and abstract concepts known to them in their original home. Thus there are the root-words for ox (-nombe, -ombe, -nte), goat (-budi, -buzi, -buri), pig (-guluba), pigeon (-jiba), buffalo (nyati), dog (mbwa), hippopotamus (-bugu, gubu), elephant (-jobo, -joko), leopard (ngwi), house (-zo, -do, -yumba, -anda, -dago, -dabo), moon (-ezi), sun, sky, or God (-juba), water (-ndi, -ndiba, mandiba), lake or river (-anza),[4] drum (ngoma), name (-ina or jina), wizard (nganga), belly, bowel (-vu, -vumo), buttocks (-tako); adjectives like -bi (bad), -eru (white); the numerals, 2, 3, 4, 5, 10 and 100; verbs like fwa (to die), ta (to strike, kill), la (da) or lia (di, dia) (to eat). The root-words cited are not a hundredth part of the total number of root-words which are practically common to all the spoken dialects of Bantu Africa. Therefore the possession amongst its root-words of a common name for "fowl" seems to the present writer to show conclusively that (1) the original Bantu tribe must have possessed the domestic fowl before its dispersal through the southern half of Africa began, and that (2) as it is historically certain that the fowl as a domestic bird did not reach Egypt before the Persian conquest in 525 B.C., and probably would not have been transmitted to the heart of Africa for another couple of hundred years, the Bantu exodus (at any rate to the south of the equatorial region) may safely be placed at a date not much anterior to 2100 years ago.

The creation of the Bantu type of language (pronominal-prefix) was certainly a much more ancient event than the exodus from the Bantu mother-land. Some form of speech like Fula, Kiama (Tern), or Kposo of northern Togoland, or one of the languages of the lower Niger or Benue, may have been taken up by ancient Libyan, Hamite or Nilotic conquerors and cast into the type which we now know as Bantu,—a division of sexless Negro speech, however, that shows no obvious traces of Hamitic (Caucasian) influence. We have no clue at present to the exact birth-place of the Bantu nor to the particular group of dialects or languages from which it sprang. Its origin and near relationships are as much a puzzle as is the case with the Aryan speech. Perhaps in grammatical construction (suffixes taking the place of prefixes) Fula shows some resemblance; and Fula possesses the concord in a form considerably like that of the Bantu, as well as offering affinities in the numerals 3 and 4, and in a few nominal, pronominal and verbal roots. The Timne and cognate languages of Sierra Leone and the north Guinea coast use pronominal prefixes and a system of concord, the employment of the latter being precisely similar to the same practice in the Bantu languages; but in word-roots (substantives, numerals, pronouns, verbs) there is absolutely no resemblance with this north Guinea group of prefix-using languages. In the numerals 2, 3, 4, and sometimes 5, and in a few verbal roots, there is a distinct affinity between Bantu and the languages of N. Togoland, the Benue river, lower Niger, Calabar and Gold Coast. The same thing may be said with less emphasis about the Madi and possibly the Nyam-Nyam (Makarka) group of languages in Central Africa though in none of these forms of speech is there any trace of the concord. Prefixes of a simple kind are used in the tongues of Ashanti, N. Togoland, lower Niger and eastern Niger delta, Cross River and Benue, to express differences between singular and plural, and also the quality of the noun; but they do not correspond to those of the Bantu type, though they sometimes fall into "classes." In the north-west of the Bantu field, in the region between Cameroon and the north-western basin of the Congo, the Cross river and the Benue, there is an area of great extent occupied by languages of a "semi-Bantu" character, such as Nki, Mbudikum, Akpa, Mbe, Bayon, Manyan, Bafut and Banshō, and the Munshi, Jarawa, Kororofa, Kamuku and Gbari of the central and western Benue basin. The resemblances to the Bantu in certain word-roots are of an obvious nature; and prefixes in a very simple form are generally used for singular and plural, but the rest of the concord is very doubtful. Here, however, we have the nearest relations of the Bantu, so far as [v.03 p.0358] etymology of word-roots is concerned. Further evidence of slight etymological and even grammatical relationships may be traced as far west as the lower Niger and northern and western Gold Coast languages (and, in some word-roots, the Mandingo group). The Fula language would offer some grammatical resemblance if its suffixes were turned into prefixes (a change which has actually taken place in the reverse direction in the English language between its former Teutonic and its modern Romanized conditions; cf. "offset" and "set-off," "upstanding" and "standing-up").

The legends and traditions of the Bantu peoples themselves invariably point to a northern origin, and a period, not wholly removed from their racial remembrance, when they were strangers in their present lands. Seemingly the Bantu, somewhat early in their migration down the east coast, took to the sea, and not merely occupied the islands of Pemba and Zanzibar, but travelled as far afield as the Comoro archipelago and even the west coast of Madagascar. Their invasion of Madagascar must have been fairly considerable in numbers, and they doubtless gave rise to the race of black people known traditionally to the Hovas as the Va-zimba.

The accompanying map will show pretty accurately the distribution of the Bantu-speaking Negroes at the present day.



It will be seen by a glance at this map that the areas in which are spoken Bantu languages of typical structure and archaic form are somewhat widely spread. Perhaps on the whole the most archaic dialects at the present day are those of Mount Elgon, Ruwenzori, Unyoro, Uganda, the north coast of Tanganyika and of the Bemba country to the south-west of Tanganyika; also those in the vicinity of Lake Bangweulu, and the Nkonde and Kese dialects of the north and north-east coasts of Lake Nyasa; also (markedly) the Subiya speech of the western Zambezi. Another language containing a good many original Bantu roots and typical features is the well-known Oci-herero of Damaraland (though this S.W. African group also presents marked peculiarities and some strange divergencies). Kimakonde, on the east coast of Africa, is a primitive Bantu tongue; so in its roots, but not in its prefixes, is the celebrated Ki-swahili of Zanzibar. Ci-bodzo of the Zambezi delta is also an archaic type of great interest. The Zulu-Kaffir language, though it exhibits marked changes and deviations in vocabulary and phonetics (both probably of recent date), preserves a few characteristics of the hypothetical mother-tongue: so much so that, until the languages of the Great Lakes came to be known, Zulu-Kaffir was regarded as the most archaic type of Bantu speech, a position from which it is now completely deposed. It is in some features unusually divergent from the typical Bantu.

Classification.—With our present knowledge of the existing Bantu tongues and their affinities, it is possible to divide them approximately into the following numbered groups and subdivisions, commencing at the north-eastern extremity of the Bantu domain, where, on the whole, the languages approximate nearest to the hypothetical parent speech.

(1) The Uganda-Unyoro group. This includes all the dialects between the Victoria Nile and Busoga on the east and north, the east coast of Lake Albert, the range of Ruwenzori and the Congo Forest on the west; on the south-east and south, the south coast of the Victoria Nyanza, and a line from near Emin Pasha Gulf to the Malagarazi river and the east coast of Tanganyika. On the south-west this district is bounded more or less by the Rusizi river down to Tanganyika. It includes the district of Busoga on the north-east and all the archipelagoes and inhabited islands of the Victoria Nyanza even as far east as Bukerebe, except those islands near the north-east coast. The dialects of Busoga, the Sese Islands and the west coast of Lake Victoria are closely related to the language of the kingdom of Uganda. Allied to, yet quite distinct from the Uganda subjection, is that which is usually classified as Unyoro.[5] This includes the dialects spoken by the Hima (Hamitic aristocracy of these equatorial lands—Uru-hima, Ru-hinda, &c.), Ru-songora, Ru-iro, Ru-toro, Ru-tusi, and all the kindred dialects of Karagwe, Busiba, Ruanda, Businja and Bukerebe. Ki-rundi, of the Burundi country at the north end of Tanganyika, and the other languages of eastern Tanganyika down to Ufipa are closely allied to the Unyoro sub-section of group 1, but perhaps adhere more closely to group 12. The third independent sub-section of this group is Lu-konjo, the language which is spoken on the southern flanks of the Ruwenzori Range and thence southwards to Lake Kivu and the eastern limits of the Congo Forest.

(2) The second group on the geographical list is Lihuku-Kuamba, the separate and somewhat peculiar Bantu dialects lingering in the lands to the south and south-west of Albert Nyanza (Mboga country). Lihuku (or Libvanuma) is a very isolated type of Bantu, quite apart from the Uganda-Unyoro groups, with which it shows no special affinity at all, though in close juxtaposition. Its alliance with Kuamba of western Ruwenzori is not very close. Other affinities are with the degraded Bantu dialects (Ki-bira, &c.) of the Ituri-Aruwimi forests. Kuamba is spoken on the west and north slopes of Ruwenzori. Both Kuamba and Lihuku show a marked relationship with the languages on the northern Congo and Aruwimi, less in grammar than in vocabulary.

(3) The Kavirondo-Masaba section. This group, which includes the Lu-nyara, Luwanga, Lukonde and Igizii of the north-east and eastern shores of the Victoria Nyanza and the northern Kavirondo and Mount Elgon territories, is related to the Luganda section more than to any group of the Bantu tongues, but it is a very distinct division, in its prefixes the most archaic. It includes the languages spoken along the western flanks of Mount Elgon, those of Bantu Kavirondo, and of the eastern coast-lands of the Victoria Nyanza (Igizii).

(4) The Kikuyu-Kamba group of British East Africa, east of the Rift valley. It includes, besides the special dialects of Kikuyu and Ukambani, all the scattered fragments of Bantu speech on Mount Kenya and the upper Tana river (Dhaicho).

(5) The Kilimanjaro (Chaga-Siha) group, embracing the rather peculiar dialects of Mount Kilimanjaro, Mount Meru and Ugweno.

(6) The Pokomo-Nyika-Giriama-Taveita group represents the Bantu dialects of the coast province of British East Africa, between (and including) the Tana river on the north and the frontier of German East Africa on the south.

(7) Swahili, the language of Zanzibar and of the opposite coast, a form of speech now widely spread as a commercial language over Eastern and Central Africa. Swahili is a somewhat archaic Bantu dialect, indigenous probably to the East African coast south of the Ruvu (Pangani) river, which by intermixture with Arabic has become the lingua franca of eastern Africa between the White Nile and the Zambezi. It was almost certainly of mainland origin, distinct from the original local dialects of Zanzibar and Pemba, which may have belonged to group No. 6. There are colonies of Swahili-speaking people at Mombasa, Malindi, Lamu, and even as far north as the Shebeli river in Somaliland, also along the coast of German and Portuguese East Africa as far south as Angoche. In the coast-lands between the Ruvu or Pangani river on the north and the Kilwa settlements on the south, the local languages and dialects are more or less related to Swahili, though they are independent languages. Amongst these may be mentioned Bondei, Shambala (north of the Ruvu), Nguru, Zeguha, Ki-mrima and Ki-zaramo.

(8) This group might be described as Kaguru-Sagala-Kami. It is one which occupies the inland territories of German East Africa, between the Swahili coast dialects on the east and the domain of the Nyamwezi (No. 11) on the west. On the north this group is bounded by the non-Bantu languages of the Masai, Mbugu and Taturu, and on the south by the Ruaha river. This group includes Kigogo and Irangi.

[v.03 p.0359] (9) The dialects of the Comoro Islands, between the East African coast and Madagascar, are styled Hi-nzua or Anzuani and Shi-ngazija. They are somewhat closely related to Swahili.

(10) The archaic Makonde or Mabiha of the lower Ruvuma, and the coast between Lindi and Ibo; this might conceivably be attached to the Swahili branch.

(11) The Nyamwezi group includes all the dialects of the Nyamwezi country west of Ugogo as far north as the Victoria Nyanza (where the tongues melt into group No. 1), and bounded on the south by the Upper Ruaha river, and on the west by the eastern borderlands of Tanganyika. The Nyamwezi genus penetrates south-west to within a short distance of Lake Rukwa. A language of this group was at one time a good deal spoken in the southern part of the Belgian Congo, having been imported there by traders who made themselves chiefs.

(12) The Tanganyika languages (Ki-rega, Kabwari, Kiguha, &c). These dialects are chiefly spoken in the regions west-north-west, and perhaps north and east of Tanganyika, from the vicinity of Lake Albert Edward on the north and the Lukuga outlet of Tanganyika on the south. On the west they are bounded by the Congo Forest and the Manyema genus (No. 13). The languages on the east coast of Tanganyika (Ki-rundi, Kigeye, &c.) seem to be more nearly connected with those of group No. 1 (Uganda-Unyoro), yet perhaps they are more conveniently included here.

(13) The Manyema (Baenya) group includes most of the corrupt Bantu dialects between the western watershed of Tanganyika and the main stream of the Luapula-Congo, extending also still farther north, and comprising (seemingly) the languages of the Aruwimi basin, such as Yalulema, Soko, Lokele, Kusu, Tu-rumbu, &c. On the west the Manyema group is bounded by the languages of the Lomami valley, which belong to groups Nos. 15 and 16; on the east the Manyema genus merges into the much purer Bantu dialects of groups Nos. 1 and 12. An examination of the Lihuku-Kuamba section (No. 2) shows these tongues to be connected with the Manyema group. The Kibira dialects of the north-eastern Congo Forest (Ituri district) may perhaps be placed in this section.[6]

(14) The Rua-Luba-Lunda-Marungu group (in which are included Kanyoka, Lulua and Ki-tabwa) occupies a good deal of the south central basin of the Congo, between the south-west coast-line of Tanganyika on the east and the main streams of the Kasai and Kwango on the west, between the Bakuba country on the north and the Zambezi watershed on the south.

(15) The Bakuba assemblage of Central Congo dialects (Songe, Shilange, Babuma, &c.) probably includes all the Bantu languages between the Lomami river on the east and the Kwa-Kasai and Upper Kwilu on the west. Its boundary on the north is perhaps the Sankuru river.

(16) The Balolo group consists of all the languages of the Northern Congo bend (bounded on the north, east and west by the main stream of the Congo), and perhaps the corrupt dialects of the Northern Kasai, Kwilu and Kwango (Babuma, Bahuana, Bambala, Ba-yaka, Bakutu, &c.), where these are not nearer allied to Teke (No. 18) or to Bakuba.

(17) The Bangala-Bobangi-Liboko group comprises the commercial languages of the Upper Congo (Ngala, Bangi, Liboko, Poto, Ngombe, Yanzi, &c.) and all the known Congo dialects along and to the north and sometimes south of the main stream, from as far west as the junction of the Sanga to as far east as the Rubi and Lomami rivers, and those between the Congo and the Lower Ubangi river and up the Ubangi, as far north as the limits of the Bantu domain (about 3deg 30' N.). Allied to these perhaps are the scarcely-known forms of speech in the basin of the Sanga river, besides the "Ba-yanzi" dialects of Lakes Mantumba and Leopold II.

(18) The Bateke (Batio) group. This may be taken roughly to include most of the Bantu dialects west of the Sanga river, northwest of the Lower Congo, south of the Upper Ogowe and Ngoko rivers and east of the Atlantic coast-lands.

(19) The Di-Kele and Benga dialects of Spanish Guinea and the Batanga coast of German Cameroon.

(20) The Fan or Pangwe forms of speech (so corrupt as to be only just recognizable as Bantu), which occupy the little-known interior of German Cameroon and French Gabun, down to the Ogowe, and as far east and north as the Sanga, Sanaga and Mbam rivers, and the immediate hinterland of the "Duala" Cameroon.

(21) The Duala group, which on the other hand is of a much purer Bantu type, includes the languages spoken on the estuary and delta of the Cameroon river.

(22) The Isubu-Bakwiri group of the coast-lands north of Cameroon delta (Ambas Bay), and on the west slopes of Cameroon Mts.

(23) The Bantu dialects of Fernando Po (Ediya, Bateti, Bani, &c.) distantly allied to Nos 24, 2 and 13.

(24) The Barondo-Bakundu group, which begins on the north at the Rio del Rey on the extremity of the Bantu field, near the estuary of the Cross river. This group may also include Barombi and Basā, Bonken, Abo, Nkosi and other much-debased dialects, which are spoken on the eastern slopes of the Cameroon mountains and on the Cameroon river (Magombe), and thence to the Sanaga and Nyong rivers. Eastwards and north-eastwards of this group, the languages (such as Mbe, Bati, Nki, Mbudikum, Bafut, Bayon) may be described as "semi-Bantu," and evincing affinities with the forms of speech in the basin of the Central Benue river and also with the Fan (No. 20).

(25) Turning southwards again from the north-westernmost limit of the Bantu, we meet with another group, the Mpongwe-Orungu and Aduma languages of French Gabun, and the tongues of the Lower Ogowe and Fernan Vaz promontory.

(26) These again shade on the south into the group of Kakongo dialects of the Loango and Sete Kama coast—such as Ba-kama, Ba-nyanga, Ma-yombe, Ba-vili, Ba-kamba and Ka-kongo (Kabinda).

(27) The Kongo language group comprises the dialects along the lower course of the Congo from its mouth to Stanley Pool; also the territory of the old kingdom of Congo, lying to the south of that river (and north of the river Loje) from the coast eastwards to the watershed of the river Kwango (and the longitude, more or less, of Stanley Pool).

(28) In the south the Kongo dialects melt imperceptibly into the closely-allied Angola language. This group may be styled in a general way Mbundu, and it includes the languages of Central Angola, such as Ki-mbundu, Mbamba, Ki-sama, Songo, U-mbangala. The boundary of this genus on the east is probably the Kwango river, beyond which the Lunda languages begin (No. 14). On the north, the river Loje to some extent serves as a frontier between the Kongo and Mbundu tongues. On the south the boundary of group No. 28 is approximately the 11th degree of south latitude.

(29) Very distinct from the Ki-mbundu speech (though with connecting forms) is the Oci-herero group, which includes the Herero language of Damaraland, the Umbundu of the Bihe highlands of south Angola, the Nano of the Benguela coast, and Si-ndonga, Ku-anyama and Oci-mbo of the southern regions of Portuguese Angola and the northern half of German South-West Africa. The languages of group No. 29 probably extend as far inland as the Kwito and Kubango rivers, in short, to the Zambezi watershed. On the south they are confronted with the Hottentot languages. The Haukoin or Hill Damaras—a Negro race of unexplained affinities and apparently speaking a Hottentot language—occupy an enclave in the area of Herero speech.

(30) What may be called the Kiboko or Kibokwe (also Kioko) family of eastern Angola is a language-group which seems to offer affinities to the languages of the Upper Zambezi and to those of groups Nos. 28 and 29. It extends eastwards into the south-western portion of the Belgian Congo, and includes the Lubale of northern Barotseland and the sources of the river Zambezi, and possibly the Gangela of south-western Angola.

(31) Southwards of group No. 30 is that of the Barotseland languages, of which the best-known form—almost the only one that is effectively illustrated—is Si-luyi. To Si-luyi may be related the Mabunda of Western Barotseland. The dialects of the Ambwela, A-mbwe, Ma-bukushu and A-kwamashi are probably closely related.

(32) Next is a group which might be styled the Subiya-Tonga-Ila, though some authorities think that Tonga and Ila deserve to be ranked as an independent group. There is, however, a close alliance in structure between the languages of each of the two subsections. The Tonga subgroup would include the dialects of the Ba-tetela, the Ba-ila (Mashukulumbwe) and of all Central Zambezia. Ci-subiya is the dominant language of South-West Zambezia, along a portion of the Zambezi river south of Barotseland, and in the lands lying between the Zambezi and the Chobe-Linyante river. Subiya is one of the most archaic of Bantu languages, more so than Tonga. Both are without any strong affinity to Oci-herero, and only evince a slight relationship with the Zulu group (No. 44).

(33) The Bisa or Wisa family includes the languages of Iramba, Bausi, Lukinga, in the southernmost projection of the Belgian Congo, and the dialects of Lubisa and Ilala between the Chambezi river and Lake Bangweulu on the north, and the Luangwa river on the east and south; perhaps also some of the languages along the course of the Upper Luapula river.

(34) With it is closely allied that of the Bemba or Emba dialects. This interesting genus occupies the ground between the south-west and south coasts of Tanganyika, Lake Mweru, and the Upper Chambezi river. The Ki-bemba domain may be taken to include the locally-modified Ki-lungu and Ki-mambwe of South and South-East Tanganyika.

(35) What may be called the North Nyasa or Nkonde group comprises all the dialects of the north-west and north coasts of Lake Nyasa (such as Ici-wandia and Iki-nyikiusa) and Ishi-nyi[chi]a of the Nyasa-Tanganyika plateau, and extends perhaps as far north west as the Fipa country (Iki-fipa), and the shores of Lake Rukwa (Ici-wungu) in the vicinity of the Nyamwezi domain (No. 11). Iki-fipa, however, has some affinities to the Tanganyika and western Victoria-Nyanza languages (groups Nos. 1 and 12).

[v.03 p.0360] (36) The western part of Nyasaland, south of group No. 35, is occupied by the Tumbuka section, which includes the languages of the Tumbuka, Henga and A-tonga peoples, and occupies the area between the western shores of Lake Nyasa and the Upper Luangwa river.

(37) Eastwards of No. 35 (North Nyasa group) lies the Kinga speech of the lofty Livingstone mountains, which is sufficiently distinct from its neighbours to be classified as a separate group.

(38) East of the Livingstone mountains and west of the Ruaha river, south also of the Unyamwezi domain, extends the Sango-Bena-Hehe-Sutu group.

(39) The extensive Yao genus of languages stretches from just behind the coast of the Lindi settlements in German East Africa (Ki-mwera) south-westward across the Ruvuma river to the north-east shores of Lake Nyasa (Ki-kese), and thence back to the valley of the Lujenda-Ruvuma (Cingindo), and southwards in various dialects of the Yao language to the south-east corner of Lake Nyasa and the region east of the Shire river, between Lake Nyasa, the Shire highlands and Mt. Mlanje. It is only since the middle of the 19th century that the Yao language has conquered territory to the south of Lake Nyasa. There still remain within its domain colonies of Nyanja-speaking people.

(40) Eastwards of the Yao domain, and bounded on the north by the range of that language in the Ruvuma valley and by the separate group of Ki-makonde (No. 10), ranges the well-marked Makua genus. The languages thus described occupy the greater part of Portuguese East Africa away from the watershed of Lake Nyasa. The Makua language is probably divided into the following dialects:—I-medo, I-lomwe, I-tugulu and Anguru. There are other dialects unnamed in the Angoji coast-region, where, however, strong colonies of Swahili-speaking people are settled. The southern part of the Makua domain is occupied by the Ci-cuambo of the Quelimane district.

(41) Nyanja, perhaps the most extensive group of cognate languages in the Bantu field, is principally associated with the east and west shores of the southern half of Lake Nyasa. It also covers all the valley of the Shire, except portions of the Shire highlands, down to the junction of that stream with the Zambezi, and further, the lands on both banks of the Zambezi down to and including its delta. West of Lake Nyasa, the Nyanja domain extends in the Senga language to the river Luangwa and the Central Zambezi, also along both banks of the Central Zambezi. South of the Central Zambezi, Nyanja dialects are spoken as far west as the Victoria Falls. Thence they extend eastwards over Mashonaland to the sea-coast. With this family may also be associated the languages of the Portuguese coast-region south of the Zambezi as far as Inhambane. The principal dialects of the Nyanja language are the Cinyanja of Eastern Nyasaland, Ci-peta and Ci-maravi of South-West Nyasaland to as far as the watershed of the Luangwa river, the Ci-mananja of the Shire highlands, Ci-mobo and Ci-machinjiri of the Shire valley, Ci-sena or Ci-nyungwe of Tete and Sena (Zambezi), and Ci-mazaro of the Lower Zambezi. The Luangwa regions, as already mentioned, are occupied by the distinct but closely-allied Senga language. South of the Central Zambezi there are Ci-nanzwa in the region near the Victoria Falls, Ci-nyai, Shi-kalana, Ci-shuna (Ci-gomo), Ci-loze, and possibly Ci-shangwe (or Ci-hlangane) and Shi-lenge which link on to the Beira coast dialects. In the delta of the Zambezi is to be found Ci-podzo, a very distinct language, yet one which belongs to the Nyanja genus. Ci-shangane, Chopi or Shi-lenge and other dialects of the Beira and Inhambane coast-lands and of Manika have been much influenced by Zulu dialects (Tebele and Ronga).

(42) The well-marked Bechuana language group has very distinct features of its own. This includes all the Bantu dialects of the Bechuanaland protectorate west of the Guai river. Bechuana dialects (such as Ci-venda, Se-suto, Se-peli, Se-rolon, Se-[chi]lapi, &c.) cover a good deal of the north and west of the Transvaal, and extend over all the Orange River Colony and Bechuanaland. Se-suto is the language of Basutoland; Se-rolon, Se-mangwato, of the Eastern Kalahri; Se-kololo is the court language of Barotseland; Ci-venda and Se-pedi or Peli are the principal dialects of the Transvaal. Group No. 42, in fact, stretches between the Zambezi on the north and the Orange river on the south, and extends westward (except for Hottentot and Bushmen interruptions) to the domain of the Oci-herero.

(43) The Ronga (Tonga) languages of Portuguese South-East Africa (Gazaland, Lower Limpopo valley, and patches of the North Transvaal (Shi-gwamba), Delagoa Bay) are almost equally related to the Nyanja group (41) on the one hand, and to Zulu on the other, probably representing a mingling of the two influences, of which the latter predominates.

(44) Lastly comes the Zulu-Kaffir group, occupying parts of Rhodesia, the eastern portion of the Transvaal, Swaziland, Natal and the eastern half of Cape Colony. In vocabulary, and to some degree in phonetics, the Zulu language (divided at most into three dialects) is related in some phonetic features to No. 42, and of course to No. 43; otherwise it stands very much alone in its developments. It may have distant relations in groups Nos. 29 and 32. Dialects of Zulu (Tebele and Ki-ngoni or Ci-nongi) are spoken at the present day in South-West Rhodesia and in Western Nyasaland and on the plateaus north-east of Lake Nyasa, carried thither by the Zulu raiders of the early 19th century.

The foregoing is only an attempt to classify the known forms of Bantu speech and to give their approximate geographical limits. The writer is well aware that here and there exist small patches of languages spoken by two or three villages which, though emphatically Bantu, possess isolated characters making them not easily included within any of the above-mentioned groups; but too detailed a reference to these languages would be wearisome and perhaps puzzling. Broadly speaking, the domain of Bantu speech seems to be divided into four great sections:—(a) the languages of the Great Lakes and the East Coast down to and including the Zambezi basin; (b) the South-Central group (Bechuana-Zulu); (c) the languages of the South-West, from the southern part of the Belgian Congo to Damaraland and the Angola-Congo coast; and (d) the Western group, including all the Central and Northern Congo and Cameroon languages, and probably also group No. 2 of the Albert Nyanza and Semliki river.

Common Features.—There is no mistaking a Bantu language, which perhaps is what renders the study of this group so interesting and encouraging. The homogeneity of this family is so striking, as compared with the inexplicable confusion of tongues which reigns in Africa north of the Bantu borderland, that the close relationships of these dialects have perhaps been a little exaggerated by earlier writers.

The phonology of the Western group (d) is akin to that of the Negro languages of Western and West-Central Africa. A small portion of (b) the South-Central group (Zulu) has picked up clicks, perhaps borrowed from the Hottentots and Bushmen. Otherwise, the three groups (a), (b) and (c) are closely related in phonology, and never, except here and there on the borders of the Western group, adopt the peculiar West African combinations of kp and gb, which are so characteristic of African speech between the Upper Nile and the Guinea coast.

The following propositions may be laid down to define the special or peculiar features of the Bantu languages:—

(1) They are agglutinative in their construction, the syntax being formed by adding prefixes principally and also suffixes to the root, but no infixes (that is to say, no mutable syllable incorporated into the middle of the root-word).[7] (2) The root excepting its terminal vowel is practically unchanging, though its first or penultimate vowel or consonant may be modified in pronunciation by the preceding prefix, or the last vowel in the same way by the succeeding suffix.

(3) The vowels of the Bantu languages are always of the Italian type, and no true Bantu language includes obscure sounds like oe and ue. Each word must end in a vowel (though in some modern dialects in Eastern Equatorial, West and South Africa the terminal vowel may be elided in rapid pronunciation, or be dropped, or absorbed in the terminal consonant, generally a nasal). No two consonants can come together without an intervening vowel, except in the case of a nasal, labial or sibilant.[8] No consonant is doubled. Apparent exceptions occur to this last rule where two nasals, two r's or two d's come together through the elision of a vowel or a labial.

(4) Substantives are divided into classes or genders, indicated by the pronominal particle prefixed to the root. These prefixes are used either in a singular or in a plural sense. With the exception of the "abstract" prefix Bu (No. 14), no singular prefix can be used as a plural nor vice versa. There is a certain degree of correspondence between the singular and plural prefixes (thus No. 2 prefix serves almost invariably as a plural to No. 3; No. 8 corresponds as a plural to No. 7). The number of prefixes common to the whole group is perhaps sixteen. The pronominal particle or prefix of the noun is attached as a prefix to the roots of the adjectives, pronouns, prepositions and verbs of the sentence which are connected with the governing noun; and though in course of time these particles may differ in form from the prefix of the substantive, they were akin in origin. (This system is the "concord" of Dr Bleek.[9]) The pronominal particles, whether in nominative or accusative case, must always precede the nominal, pronominal, adjectival and verbal roots, though they often follow the auxiliary prefix-participles used in conjugating verbs,[10] and the roots of some prepositions.

[v.03 p.0361]

(5) The root of the verb is the second person singular of the imperative.

(6) No sexual gender is recognized in the pronouns and concord. Sexual gender may be indicated by a male "prefix" of varying form, often identical with a word meaning "father," while there is a feminine prefix, na or nya, connected with the root meaning "mother," or a suffix ka or kazi, indicating "wife," "female." The 1st and 2nd prefixes invariably indicate living beings and are Usually restricted to humanity.

The sixteen original prefixes of the Bantu languages are given below in the most archaic forms to be found at the present day. The still older types of these prefixes met with in one or two languages, and deduced generally by the other forms of the particle used in the syntax, are given in brackets. It is possible that some of these prefixes resulted from the combination of a demonstrative pronoun and a prefix indicating quality or number.

Old Bantu Prefixes.

Singular. Plural.

Class 1. Umu- (Ngu-mu-).[11] Class 2. Aba (Mba-ba or Nga-ba).[11] " 3. Umu- (Ngu-mu-). " 4. Imi- (Ngi-mi-). " 5. Idi (Ndi-di-). " 6. Ama- (Nga-ma-). " 7. Iki- (Nki-ki-). " 8. Ibi- (Mbi-bi-). " 9. I-n- or I-ni- (?Ngi-ni-). " 10. Iti-, Izi-, Iti-n-, Izi-n- (?Ngi-ti-). " 11. Ulu (Ndu-du-). " 12. Utu (?Ntu-tu-); often diminutive in sense. " 13. Aka (?Nka-ka-); usually diminutive, sometimes honorific. " 14. Ubu- (?Mbu-bu-); sometimes used in a plural sense; generally employed to indicate abstract nouns. " 15. Uku (?Nku-ku-); identical with the preposition "to," used as an infinitive with verbs, but also with certain nouns indicating primarily functions of the body. " 16. Apa (Mpa-pa-); locative; applied to nouns and other forms of speech to indicate place or position; identical with the adverb "here," as Ku- is with "there."

To these sixteen prefixes, the use of which is practically common to all members of the family, might perhaps be added No. 17, Fi- or Vi-, a prefix in the singular number, having a diminutive sense, which is found in some of the western and north-western Bantu tongues, chiefly in the northern half of the Congo basin and Cameroon. It is represented as far east (in the form of I-) as the Manyema language on the Upper Congo, near Tanganyika. This prefix cannot be traced to derivation from any others among the sixteen, certainly not to No. 8, as it is always used in the singular. Its corresponding plural prefix is No. 12 (Tu-). Prefix No. 18 is Ogu-, which has, as a plural prefix, No. 19, Aga-. These are both used in an augmentative sense, and their use seems to be confined to the Luganda and Masaba dialects, and perhaps some branches of the Unyoro language. These, like No. 17, are regular prefixes, since they are supplied with the concord (-gu- and -ga-). Lastly, there is the 20th prefix, Mu-, which is really a preposition meaning "in" or "into," often combined in meaning with another particle, -ni, used always as a suffix.[12] The 20th prefix, Mu-, however, does not seem to have a complete concord, as it is only used adjectivally or as a preposition and has no pronominal accusative.

The concord may be explained thus:—Let us for a moment reconstruct the original Bantu mother-tongue (as attempts are sometimes made to deduce the ancient Aryan from a comparison of the most archaic of its daughters) and propound sentences to illustrate the repetition of pronominal particles known as the concord.

Old Bantu. Babo mbaba-ntu[13] babi ba-bo-ta tu-ba-oga. They these-they person they bad they who kill we fear them. Rendered into the modern dialect of Luganda this would be:— Bo aba-ntu ba-bi babota tu-ba-tia. They these-they person they bad they who kill we them fear. (They are bad people who kill; we fear them.)

Old Bantu. Ngu-mu-ti nguno ngu-gwa ku-ngu-mbona. This tree this here this falls; thou this seest? Rendered into Kiguha of North-West Tanganyika, this would be:— Umuti guno gugwa ugumona? It tree this here it falls; thou it seest? (The tree falls; dost thou see it?)

The prefixes and their corresponding particles have varied greatly in form from the original syllables, as the various Bantu dialects became more and more corrupt. Assuming these prefixes to have consisted once of two distinct particles, such as, for example, Nos. 1 and 3, Ngu-mu-, or the 6th plural prefix Nga-ma-, the first syllable seems to have been of the nature of a demonstrative pronoun, and the second more like a numeral or an adjective. Mu- probably meant "one," and Ma- a collective numeral of indefinite number, applied to liquids (especially water), a tribe of men, a herd of beasts—anything in the mass.[14] In the corresponding particles of the concord as applied to adjectives, verbs and pronouns, sometimes the first syllable, Ngu or Nga was taken for the concord and sometimes the second mu or ma. This would account for the seemingly inexplicable lack of correspondence between the modern prefix and its accompanying particle, which so much puzzled Bleek and other early writers on the Bantu languages. In many of these tongues, for example, the particle which corresponds at the present day to the plural prefix Ma- is not always Ma, but more often Ga-, Ya-, A-; while to Mu- (Classes 1 and 3) the corresponding particle besides -mu- is gu-, gw-, u-, wu-, yu-, n-, &c.

The second prefix. Ba- or Aba-, is, in the most archaic Bantu speech (the languages of Mt. Elgon), Baba- in its definite form (Ngaba sometimes in Zulu-Kaffir). The concord is -ba- in all the less corrupt Bantu tongues, but this plural prefix degenerates into Va-, Wa-, Ma-, and A-. The concord of the 4th prefix, Mi-, is gi-, -i-, -ji-, and sometimes -mi-. The commonest form of the 5th prefix at the present day is Li- (the older and more correct is Di-), and its concord is the same; this 5th prefix is often dropped (the concord remaining) or becomes Ri-, I-, Ji-, and Ni-. The 7th prefix, Ki-, in many non-related dialects pursues a parallel course through Ci- into Si- (=Shi) and Si- and its concord resembles it. The 8th prefix is still more variable. In its oldest form this is Ibi- or Mbibi-. It is invariably the plural of the 7th. It becomes in different forms of Bantu speech Vi-, Pi-, Fi-, Fy-, Pši-, Ši-, I-, By-, Bzi-, Psi-, Zwi-, Zi- and Ri-, with a concord that is similar. The 10th prefix, which was originally Ti- or Tin-, or Zi- or Zin-, becomes Jin-, Rin-, Din-, Lin-, [theta]in-, [theta]on-. &c. The n in this prefix is really the singular prefix No. 9, which is sometimes retained in the plural, and sometimes omitted. In the case of the 10th prefix, the concord or corresponding pronoun persists long after the prefix has fallen out of use as a definite article. Thus, though it is absent as a plural prefix for nouns in the Swahili of Zanzibar, it reappears in the concord. For instance:—Nombe hizi zangu—Cows these mine (These cows are mine), although Nombe has ceased to be zinombe in the plural, the Zi- particle reappears in hizi and zangu. In fact, the persistence of this concord, which exists in almost every known Bantu language in connexion with the 10th prefix, shows that prefix to have been in universal use at one time. The 11th prefix -Lu- seems to be descended from an older form, Ndu-. Its commonest type is Lu-, but it sometimes loses the L and becomes U-, and in the more archaic dialects is usually pronounced Du- or Ru-. It is also Nu- in one or two languages. The 12th prefix (Tu-), always used in a diminutive sense, disappears in many of these languages. Where met with it is generally Tu- or To-, but sometimes the initial T becomes R (Ru-, Ro-) or L (Lu-, Lo-) or even Y (Yo-), the concord following the fortunes of the prefix. The 13th prefix (Ka-) is sometimes confused with the 7th (Ki) and merged into it and vice versa. Ka- very often takes the 8th prefix as a plural, more commonly the 12th, sometimes the 14th. This prefix (Ka-) entirely disappears in the north-western section of the Bantu languages. Bleek thought that it persisted in the attenuated form of E- so characteristic of the Cameroon and northern Congo languages, but later investigations show this E- to be a reduction of Ki- (Ke-) the 7th prefix. The 14th prefix Bu- is very persistent, but frequently loses its initial letter B, which is either softened into V or W, or disappears altogether, the prefix becoming U- or O- or Ow-. Sometimes this prefix becomes palatized into By- or even Tš- (C-). The concord follows suit. The 15th prefix, Ku-, occasionally loses its initial K or softens into Hu or [Greek: chu] or strengthens into Gu. Its concord under these circumstances sometimes remains in the form of Ku-. The 16th, Pa-, prefix is one of the most puzzling in its distribution and its phonetic changes. A very large number of the Bantu languages in the north, east and west have a dislike to the consonant P, which they frequently transmute into an aspirate (H), or soften into V, W, or F, or simply drop out. There is too much evidence in favour of this prefix having been originally Pa- or Mpa-pa to enable us to give it any other form in reconstructing the Bantu mother-tongue. Yet in the most archaic Bantu dialects to the north of the Victoria Nyanza it is nowhere found in the form of Pa-. It is either Ha- (and Ha- changes eastward into Sa-!) or Wa-.[15] But for its existence in this shape in the language of Uganda one might almost be led to think that the 16th locative prefix began as Ha-, and by some process without a parallel changed in the east and south to the form of Pa-. There are, however, a good many place names in the northern part of the Uganda protectorate, in the region now occupied by Nilotic negroes, which begin with Pa-. These place names would seem to be of ancient Bantu origin in a [v.03 p.0362] land from which the Bantu negroes were subsequently driven by Nilotic invaders from the north. They may be relics therefore of a time before the Pa- prefix of those regions had changed to the modern form of Ha-. In S.W. and N.W. Cameroon the initial p of the 16th prefix reappears in two or three dialects; but elsewhere in North-West Bantu Africa and in the whole basin of the Congo, except the extreme south and south-east, the form Pa- is never met with; it is Va-, Wa-, Ha-, Fa-, or A-. In the Secuana group of dialects it is Fa- or Ha-; in the Luyi language of Barotseland it assumes the very rare form of Ba-, while the first prefix is weakened to A-.