But the opposite fault, once more common, should be equally avoided. Words should not be used merely because they are 'old' or obsolete. The words chosen, however remote they may be from colloquial speech or ephemeral suggestions, must be words that remain in literary use, especially in the use of verse, among educated people. (To such Beowulf was addressed, into whatever hands it may since have fallen.) They must need no gloss. The fact that a word was still used by Chaucer, or by Shakespeare, or even later, gives it no claim, if it has in our time perished from literary use. Still less is translation of Beowulf a fitting occasion for the exhumation of dead words from Saxon or Norse graves. Antiquarian sentiment and philological knowingness are wholly out of place. To render leode 'freemen, people' by leeds (favoured by William Morris) fails both to translate the Old English and to recall leeds to life. The words used by the Old English poets, however honoured by long use and weighted with the associations of old verse, were emphatically those which had survived, not those which might have survived, or in antiquarian sentiment ought to have survived.

Different, though related, is the etymological fallacy. A large number of words used in Beowulf have descended to our own day. But etymological descent is of all guides to a fit choice of words the most untrustworthy: wann is not 'wan' but 'dark'; mod is not 'mood' but 'spirit' or 'pride'; burg is not a 'borough' but a 'strong place'; an ealdor is not an 'alderman' but a 'prince'. The vocabulary of Old English verse may have philological interests but it had no philological objects.[5]

The difficulties of translators are not, however, ended with the choice of a general style of diction. They have still to find word for word: to deal with the so-called 'synonyms' of Old English verse and with the compounds. Translation of the individual simple words means, or should mean, more than just indicating the general scope of their sense: for instance, contenting oneself with 'shield' alone to render Old English bord, lind, rand and scyld. The variation, the sound of different words, is a feature of the style that should to some degree be represented, even if the differences of original meaning are neglected by the poet or no longer remembered-events which in early Old English poetry probably occurred far less often than is sometimes supposed. But in cases where Old English has built up a long list of synonyms, or partial equivalents, to denote things with which Northern heroic verse was specially concerned - such as the sea, and ships, and swords, and especially men (warriors and sailors), it will sometimes be found impossible to match its richness of variation even with the most indiscriminate collection of words. For man in Beowulf there appear at least ten virtual synonyms: beorn, ceorl, freca, guma, hæleð and hæle, leod, mann and manna, rinc, secg, and wer.[6] This list can be extended to at least twenty-five items by the inclusion of words whose sense remained in varying degrees more specific, though in heroic verse they could as a rule replace the simple mann: words implying noble birth such as æðeling and eorl; meaning youths or young men, such as cniht, hyse, maga, mecg; or denoting the various companions, followers, and servants of lords and kings, such as gædeling, geneat, gesið, scealc, ðegn, or explicitly signifying 'warrior', such as cempa, oretta, wiga, wigend. With this list not even a hotch-potch series such as man, warrior, soldier, mortal, brave, noble, boy, lad, bachelor, knight, esquire, fighter, churl, hero, fellow, cove, wight, champion, guy, individual, bloke, will compete: not even in length, certainly not in fitness. In such a case (the most extreme) we have to be content with less variation - the total effect is probably not much changed: our ears, unaccustomed to this kind of thing, may be as much impressed by less. There is, however, no need to increase our poverty by avoiding words of chivalry. In the matter of armour and weapons we cannot avoid them, since our only terms for such things, now vanished, have come down through the Middle Ages, or have survived from them. There is no reason for avoiding knights, esquires, courts, and princes. The men of these legends were conceived as kings of chivalrous courts, and members of societies of noble knights, real Round Tables. If there be any danger of calling up inappropriate pictures of the Arthurian world, it is a less one than the danger of too many warriors and chiefs begetting the far more inept picture of Zulus or Red Indians. The imagination of the author of Beowulf moved upon the threshold of Christian chivalry, if indeed it had not already passed within.

The translation of the compounds sets a different problem, already glanced at above. A satisfactory solution will seldom be arrived at by translation of the elements separately and sticking them together again: for instance, by rendering the 'kenning' or descriptive compound gleo-beam 2263, denoting the harp, as 'glee-beam', or (avoiding the etymological fallacy) as 'mirth-wood'. Of brimclifu 222 an accurate and acceptable translation may be 'seacliffs', but this is a rare good fortune. A literal rendering of 815 sele hlifade heah ond homgeap, heaðowylma bad laðan liges; ne was hit lenge ða gen ðæt se ecghete aðumsweoran æfter wælniðe wæcnan scolde would be like this: 'hall towered high and horn-spacious; war-surges awaited of hostile flame; it was not at hand yet that the blade-hate of son-father-in-law after slaughter-malice should awake'. But this is certainly not modern English, even if it is intelligible.

It is plain that the translator dealing with these compounded words must hesitate between simply naming the thing denoted (so 'harp' 1065, for gomen-wudu 'play-wood'), and resolving the combination into a phrase. The former method retains the compactness of the original but loses its colour; the latter retains the colour, but even if it does not falsify or exaggerate it, it loosens and weakens the texture. Choice between the evils will vary with occasions. One may differ in detail from the present translation, but hardly (if one respects modern as well as ancient English) in general principle: a preference for resolution.

The compounds found in Old English verse are not, however, all of the same kind, and resolution is not in all cases equally desirable. Some are quite prosaic: made for the expression of ideas without poetic intention. Such words are found both in verse and prose, and their translation depends simply on their meaning as a whole. It is not necessary to 'resolve' mundbora,[7] since the simple words 'protector' or 'patron' get as near as we can to the meaning of this word.

A larger, intermediate, class is formed by those words in which composition is used as a natural and living device of the contemporary English language. The distinction between verse and prose or colloquial use here lies mainly in the fact that these compounds arc more frequent in verse, and coined with greater freedom. In themselves - even those which are only used, or at least are only recorded, in verse - they would sound as natural in contemporary ears as would tobacco-stall or tea-drinker in ours. Of this class are heals-beag 'neck-ring', bat-weard 'boat-guard', and hord-wela 'hoard(ed) wealth' - three examples which (probably by mere chance) only occur in Beowulf, No 'Anglo-Saxon' who heard or read them would have been conscious that they were combinations never before used, even if he had in fact never met them before. Our language has not lost, though it has much limited, the compounding habit. Neither 'neck-ring' nor 'boat-guard' are recorded in the Oxford Dictionary,[8] but they are inoffensive, although 'hoard-wealth' is now unnatural. This class of compound is in general the one for which compound equivalents in modern English can with discretion most often be found or made.