The surveys so far agree with the Codes. The villata of the surveys was the taxable unit, and in some cases still paid the tunc pound (or 20s.) in lieu of the chieftain’s food-rents. In other cases escheats and other causes had varied the amount. In the Codes of South Wales the unit for the tunc pound was the tref, and in the Venedotian Code of North Wales the maenol of four trefs.

Now, as in the surveys the family groups or gwelys were located so as to occupy sometimes several villatæ, and sometimes undivided shares in villatæ along with others, so, if we may take the villata of the surveys as equivalent to the tref or maenol of the Codes, we must expect to find that the kindreds of tribesmen at the period of the Codes were scattered in the same way over the trefs and maenols. And, as the maenol was a group of trefs, the tref is the unit of tribal occupation as to which a clear understanding is most necessary. In this, however, we may be, after all, only partly successful.

The word tref, though generally used for a homestead or hamlet, seems from its other meanings to involve the idea of a group.

There were cases in which a disputed matter of fact had to be established upon the evidence of men of the gorvotref, i.e. by men of the groups outside the tref in which the question in dispute arose.[30] And this gorvotref was not merely the next adjoining tref or trefs, but it consisted of those randirs or divisions of neighbouring trefs of uchelwrs, or tribesmen, whose boundaries touched the tref in which the disputed facts arose. Neighbouring randirs of taeog trefs, i.e. the trefs of non-tribesmen, were excluded, presumably because the testimony of taeogs in matters relating to tribesmen was not relied on. But this compound of the word tref implies that its general sense was a group of homesteads. That, in general, trefs had defined boundaries, is clear from the fact that it was an offence to break them, and this applied also to the randirs or divisions of the tref.[31]

Speaking, then, of the group generally known as a tref, we must regard it, not only as a taxable area, but also as the natural group known everywhere as a trefgordd, i.e. the natural group of the homesteads of relatives or neighbours acting together as a single community as regards their cattle and their ploughing.

The typical lawful trefgordd is thus described:—

There is another passage which mentions the nine buildings in the tref.

So, in case of fire from negligence in a tref, the holder of the house in which it arose was to pay for the damage to the next houses on each side if they took fire.[34] And again no indemnity was to be paid to the owners in a trefgordd for damages from the fire of a smithy if covered with shingles or tiles or sods, nor from the fire of a bath, provided always that the smithy and the bath were at least seven fathoms from the other houses in the trefgordd.[35]

The description above quoted of the normal trefgordd suggests that the herd under the one herdsman did not belong to one person or homestead, but to many; and so far it seems to be consistent with the surveys which represent the villatæ as occupied by the cattle of several family groups who had grazing rights therein.

And this, too, accords with what the Denbigh Extent tells us of the individual tribesmen, viz. that only some of them had homesteads. So-and-so ‘habet domum’ or ‘non habet domum.’[36] The young tribesman with his da thus may have joined in a common homestead with some one else—probably with his parents or near relatives.

Distinguishing, then, the tref as a taxable area from the trefgordd, and still confining attention to the trefgordd as a cluster of homesteads united for the practical purpose of occupation, let us recur to the things which bound the trefgordd into one group, viz. the one plough, the one oven, the one churn, the one bull, and the one herdsman.

Here are the two elements combined of pastoral and agricultural co-operation, and the trefgordd is the local and physical unit of this co-operation.

Taking first the pastoral element, the trefgordd was a working unit of co-operative dairy-farming. The cattle of several households or individuals were put together in a common herd with a common bull and under the care of a common herdsman (bugeil) and his dog. It may be regarded as a group of the homesteads of the persons in charge of such a herd, and the tribesmen of a gwely may have cattle in the herds of more than one trefgordd.

Three things were ‘ornamental’ to a trefgordd, ‘a book, a teacher versed in song, and a smith (gov) in his smithy;’ but a trefgordd herdsman was an ‘indispensable’ of the hendrev,[37] and, when engaged with his herd in summer on the mountain, his ‘three indispensables’ were ‘a bothy, his herdsman’s dog, and a knife;’ and the three indispensables of his bothy were a roof-tree, roof-supporting forks, and wattling, and he was at liberty to cut them in any wild wood he pleased.[38]

So far, then, as the pastoral element was concerned, the trefgordd was occupied by a little group of tribesmen engaged in dairy-farming having charge of cattle in a common herd, with a common bull, and under the care of a common herdsman and his dog.

Custom, grown out of traditional experience of what a single herdsman and his dog could manage, had determined, it seems, the size of the normal herd. Thus in the Gwentian Code[39] we are told that ‘a legal herd of cattle is 24 kine.’ And custom tenaciously adhered to tribal rules in such matters.

Thus in the Denbigh Extent it is mentioned that the whole villata of Arquedelok was in manu domini by reason of escheats and exchanges, and that a portion of it was let ad firmam to nine firmarii, each of whom held for a term of years 31 acres, with one bull and 24 cows, paying per annum 73s. 4d., and rendering to the lord at the end of his term the said bull and cows or their price, together with the land and a house built thereon.[40] Here, even in a case in which Henry de Lacy was introducing into Wales holdings and herds in severalty, and very possibly introducing English tenants, he adhered to the Welsh tribal rule of the one bull and 24 cows to the herd. So also in the survey of St. David’s, under the head Glaston in Breconshire, the number 24 of grossa animalia is spoken of as the usual number ab antiqua consuetudine, and in the arrangement of common pasture one great animal is said to count as equal to twelve sheep.

The normal herd of the trefgordd was then 24 cows, or their equivalent in bullocks and sheep.

During the summer months the herdsman living out on the mountains was responsible with his dog for the cattle of the trefgordd. And his dog was worth as much as a cow or an ox, if it was one that ‘will go before the herd in the morning and behind them in the evening, and make three turns round them in the night.’[41]

Having no cattle of his own in the herd, the herdsman’s testimony as to whose cattle were injured, and as to whose cattle had done the injury, was held, when such cases arose, to be sufficient to make the owner responsible, while as regards injuries done by the cattle of one trefgordd to those of another there was joint responsibility.[42] There is common sense in such rules to begin with, and then, having grown into custom, they become perpetuated when custom is codified.

The trefgordd possessed further a common churn. This implies that the milk of the cows was thrown altogether into this one churn as in Swiss mountain communes now. One of the dues from a taeog trev, i.e. a group of non-tribesmen, was a cheese made from a day’s milking of all the cows in the herd. So that we note in passing that the taeog-tref, i.e. of non-tribesmen, also had its herd and was in fact a trefgordd.[43]

In winter the cattle came down into the lowlands and grazed on the pastures near the tyddyns or homesteads of the tref, and as each of these had its corn and cattle-yard,[44] we may conclude that each owner penned in his own cattle at night during the winter months or joined with some other tribesmen who had a homestead in doing so. The rules as to the divisions of the tyddyns probably referred to these winter homesteads so held in quasi-severalty.

We need not dwell upon the common oven. Every hamlet in Brittany possesses its common oven to this day, often in the middle of the village green. Nor need we more than mention the common plough, to the team of which the tribesmen contributed oxen for the cyvar or common ploughing of the portion of the waste agreed upon for each year’s corn crop.

The attempt to realise what this practical unit—the trefgordd—was, will not be thrown away if it should help us to understand how easily it lent itself to the arrangement of the chieftain’s food-rents or tribute in after-times of taxation. Granted that some such system of trefgordds or clusters of trefgordds pretty generally prevailed, having grown up as a matter of convenience in a grazing community, it is obvious how easily it might become the unit of tribute or taxation. Just as in the Domesday Survey the number of ploughs affords such a unit, so in a tribal community a district might easily be fiscally estimated at so many herds, or so many churns, or so many ploughs. All these would mean so many trefgordds. And whatever the relations of the trefgordd to the villata of the surveys might be, and however much or often the actual residents, with their herds, might be shifted from one district to another, the district, as in the Denbigh Extent, would remain the permanent unit for payments.

In the early stages of tribal life, when the chieftain of the tribe moved from one district to another and received his food-rents in the actual form of ‘the night’s entertainment,’ each customary place of encampment in his annual progress would become the centre at which the food-rents would be paid and services rendered for as many nights’ entertainment as his accustomed stay in the place. In later stages, when the chieftain’s dues were commuted into money, the ‘tunc pound’ in lieu of food-rents easily became, as we find it in the surveys, a charge on the district rather than on the shifting tribesmen and their herds.

And when the power of the chieftain had grown with time, and instead of ‘nights’ entertainments’ obtained in the primitive way by the actual movement of himself and his retinue from place to place, the food-rents or the tunc pounds in lieu of them were delivered at his palace, he would become the recipient of a regular revenue. And out of this revenue it would become easy for him to reward a follower or endow a church by the transfer of so many food-rents or tunc pounds in lieu of them, or the revenue from such and such a district, or of so many of its trefgordds, without disturbing the internal working of the system or the daily life of the tribesmen and their herds. When Beowulf returns to his chieftain after his exploit and is rewarded by the gift of a palace and so many ‘thousands,’ we naturally ask of what, and how it could be done. We may not be able to say off-hand what the unit was, but we get from the Welsh example some rough idea of what tribal tribute and income were, and how these could be readily gathered and transferred.

V. THE METHOD OF PAYMENT OF GALANAS BETWEEN KINDREDS.

Postponing for a while the consideration of the position of the various classes of non-tribesmen, but still keeping in view the fact that in considerable numbers they were practically sharers with the tribesmen in the rights of grazing and occupation of land, we are now in the position to realise to some extent what happened when a murder had taken place.

If it was of some one within the kindred, there was, as we have said, no slaying of the murderer. Whether it were a parricide or a fratricide, or the murder of a near kinsman, under Cymric custom there was no galanas, nothing but execration and ignominious exile.

But if a tribesman of one kindred were killed by a tribesman of another kindred, then it was a serious matter of blood feud between the kindreds, or of the payment of the blood fine. The tribal conscience demanded vengeance or composition.

It sometimes happened that the murderer had fled to a church for safety, taking his cattle with him. For the clergy or monks at the place of refuge had a herd of cattle of their own, and with them the murderer’s cattle were allowed to wander and graze so long as they returned nightly to the refuge.[45]

There he remained presumably till the kindred of the murdered tribesman, through negotiation and arrangement of the chiefs of the kindreds, had agreed to accept the payment of the galanas, if it were the case of an uchelwr or full tribesman, of 126 cows. Six cows, as we shall see hereafter, were saraad for the insult, and 120 cows galanas for the murder. The saraad was paid first—six cows or other cattle to the same value belonging to the murderer were driven from the herd in payment.

The murderer’s life was then safe, and presumably he might return with his cattle to his place.

Within a fortnight, the tribesmen of the murderer’s kindred met to apportion the payment of the rest. They came from trefgordds far and near, from the territories sometimes of various higher territorial chieftains within whose districts they had grazing rights.

The collected tribesmen having apportioned the payment, fortnight after fortnight instalments must be paid till the whole number in value of 120 cows was completed.[46]

But by whom was the payment to be made?[47]

Forty cows must first be found by the murderer, his father, mother, brothers, and sisters with him. They doubtless helped one another, but theoretically, in one or other of the common herds, there must have been cattle belonging to the murderer, his father, mother, brothers, and sisters, or how could they have paid their shares? There was nothing unreal in this liability of each to pay a share, for had the murderer been slain each one of them would have received, instead of having to pay, a share in 40 cows.

The murderer himself had to pay a third of the 40 cows if he had them. His father and mother between them paid the next third, and the brothers and sisters the remaining third, the sisters paying half what the brothers did.[48] The herds of many a trefgordd must be thinned before this could be done.

The remainder of the galanas, viz. 80 cows, fell on the kindred, to the seventh degree or fifth cousins. The paternal relations had to find two thirds of it and the maternal one third, and these kindreds embraced the descendants from the great-grandparents of the great-grandparents on both sides.

In the first fortnight the kindred on the father’s side had to find half what was due from them. In the second fortnight they had to find the other half, and in the third fortnight the maternal kindred had to find their share, till so at last the full tale of the 120 cows was paid. The oath of peace from the kindreds of the murdered man could then be given, and the murderer and his kinsmen, be at peace.[49]

But what happened if the murderer could not find the cattle for his third of the 40 cows which he and his immediate family had to find? He had yet a right, as a member of the greater kindred, to claim in aid a ‘spear penny’ from all those male kinsmen descended from a common ancestor on his father’s side two steps further back, i.e. still more distantly related to him than those included in the kindred to the seventh degree who had already paid their share. Even if the slayer were a woman, she had the same right of spear penny from the men of her kindred to help her to make her payment.[50]

So this attempt to realise what was involved in the payment of an ordinary case of galanas brings us back to the recognition of the double aspect of the kindred in the structure of tribal society—its solidarity and joint responsibility, on the one hand, as against outsiders, the whole kindred being responsible in the last resort; on the other hand the individual responsibility of its members, graduated according to nearness of relationship, for the crimes of their relative.

In Cymric tribal society this was made possible by the broad fact that both males and females in the group of kindred, on both paternal and maternal sides, liable to pay, had cattle of their own in the common herd, each having received his or her da for maintenance by right of kin and descent from the common ancestor or chieftain of the kindred. The two things surely hang together. And therefore, if we find in the laws of other tribes somewhat similar rules regarding the payment of wergelds, it probably will be worth while to inquire further whether the corresponding structure of tribal society, or something more or less equivalent to it, may not be present also.

VI. THE AMOUNT OF THE CYMRIC GALANAS.

In all the Welsh Codes the galanas, as already mentioned, is described in a peculiar form. It is a combination of two items, viz. the saraad, or payment for insult, and the galanas proper.

Thus the galanas of the innate boneddig, or young tribesman, accepted by the kindred as a tribesman of nine descents of Cymric blood, is described as ‘three kine and three score kine,’ that of the uchelwr or breyr as ‘six kine and six score kine.’

The explanation of this is obtained from the following passage:—

So also in the Gwentian Code:—

So also in the Venedotian Code:—

The reason why the wife has a share in the saraad and not in the galanas has already been explained. She suffers from the personal affront or insult to her slain husband and shares in the saraad. But she has no blood relationship with her husband, and only the husband’s kindred are therefore entitled to share in the galanas, as her husband’s kindred alone would have been concerned in the feud.

The saraad and the galanas were therefore separate things and subject to separate rules, though both payable on the murder of a tribesman. The galanas proper is what must be regarded in any comparison with Continental wergelds.

The real galanas of the uchelwr or breyr, apart from the saraad, was 120 cows, and that of the young innate boneddig who had received his da but had no family was 60 cows. In one of the Codes his galanas when married is said to be 80 cows.

Now in what currency was the galanas paid? Formerly, according to the Codes, all payments were made in cattle, and the galanas proper was reckoned in scores of cows.

But of what cow? How was the normal cow for practical purposes to be defined? It is a question worth answering, because we may probably take the Cymric method, of valuing the cow as a unit of currency in cattle, as at any rate suggestive of the methods generally adopted by other tribes.

According to the Venedotian Code the cow was of full normal value when in full milk and until her fifth calf.

Then the milk measure is described thus:—

In the Dimetian Code substantially the same rules are given, except that the measure of the cow’s milking is smaller.

The only difference is between the seven and the nine thumbs of diagonal measurement. Possibly there may be some error in the figures, and the measure may have been the same in both Codes.

Returning to the galanas; although it was reckoned in the Codes in scores of cows, a fixed equation had already been made between cows and silver.

The normal cow was equated in the Codes with ‘three scores of silver.’ And in the Latin version of the Dimetian Code the ‘score of silver’ is translated by ‘uncia argenti.’ The score of silver at the date of the Code was therefore an ounce of silver. So that the reckoning is the Frankish or Anglo-Saxon one of twenty pence to the ounce.

The score of pence of 32 wheat-grains would make the ounce of 640 wheat-grains: that is, the ounce of the pound of 240d., or 7680 wheat-grains—the pound in use in England after the time of Kings Offa and Alfred, and at the date of the Codes.

The galanas of the uchelwr or breyr being 120 cows, and the cow being reckoned at three scores or ounces of silver, the galanas would equal 360 scores or ounces, or thirty pounds of silver.

The ratio of gold to silver after the temporary disturbance under Charlemagne had, as we have seen, settled down again to the Imperial ratio of 1:12.

Now thirty pounds of 7680 wheat-grains equal 230,400 wheat-grains, and this number of silver wheat-grains divided by twelve equalled exactly 19,200 wheat-grains of gold. So that this Celtic galanas of the Cymric uchelwr or breyr of 120 cows, like so many Continental wergelds, was apparently exactly equal to 200 gold solidi of ninety-six wheat-grains, i.e. the heavy gold mina of Imperial standard.

VII. THE METHODS OF TREATMENT OF STRANGERS OR NON-TRIBESMEN.

Another point upon which special inquiry is made in this volume regards tribal methods of treating strangers in blood and slaves.

There is no subject requiring more careful investigation than the combination of circumstances out of which arose what is roughly called serfdom, i.e. the attachment of tenants to the land rendering services to a lord. I shall not be suspected of suggesting that tribal customs and methods were the sole factors which produced serfdom and of ignoring the influences which came from Roman methods of managing landed estates, and from Roman law modified by ecclesiastical usage.

Indeed, I have insisted from the first that while, in the ‘Germania’ of Tacitus, the germs may be found of an ‘embryo manor,’ both Roman and German elements probably combined in producing the later manorial system and serfdom which grew up in what were once the Roman provinces of Gaul and the two Germanies, and even also in Britain.[56] But I think that in Cymric tribal custom we may find a fresh clue worth following in the attempt to gather from Continental evidence the methods likely to be used by conquering German or Anglo-Saxon tribes in the treatment of strangers in blood.[57]

In Welsh tribal custom alltuds or strangers and their descendants (not necessarily otherwise unfree persons) having some special circumstances in their favour, being allowed to settle within the district of a greater or lesser chieftain upon land which, in a sense, may have been his demesne land, were free to remove and settle under another chieftain, unless and until they had remained on the same land or under the same lordship for four generations. But thereafter the great-grandchildren of the original settlers became adscripti glebæ. And this fixture to the land, or rather to the lordship, was apparently not looked upon as in any way a degradation in rank, but on the contrary a step in advance towards the recognition of tribal rights. The great-grandson of the stranger did not indeed become a Cymric tribesman, but he gained the recognition of his status as the founder of a kindred of his own, the members of which in after-generations would, as kinsmen, be able to swear for and defend one another.

This being so in the case of free strangers coming into the country, the next question is what was the position of the semi-servile class, the aillts and taeogs of the Codes, who and whose ancestors for many generations had been born upon the land in a semi-servile condition?

The fixture to the land of the aillt or taeog was not the special mark so much of a semi-servile condition as of his want of recognised kindred, and under the local custom of South Wales it seems that he too, like the alltud, could sometimes arrive at the recognition of kindred, without indeed becoming a Cymric tribesman, at the end of four generations of residence under the chieftain of the land; and even to further recognition of it, involving a still better position as to rights, at the ninth generation. The ninth man in South Wales seems according to local custom in some districts to have, at last, climbed the highest rung of the ladder, and to have attained the right to claim the status of a Cymric tribesman.

This curious rise under Cymric custom, by steps of four generations, up the ladder towards the recognition of tribal rights, seems to have a suggestive correspondence with the reverse process under manorial usage of proving the serfdom of a nativus by showing that the great-grandfather was a nativus on the lord’s land, the manorial rule being that settlement on servile land for four generations made the posterity of an original settler into nativi.[58]

Once more let us try to realise what this meant, and what was the position of these Cymric non-tribesmen in regard to their settlement on land.

If under the guidance of the Codes we turn to the extents and surveys, we find them living, in some cases, not mixed up with the tribesmen, but in separate groups, or trefs, or trefgordds. There may be here and there exceptional alltuds or strangers of a higher class growing up, by the gradual process of intermarriage for four generations with tribeswomen, into the status of tribesmen. But the mass of the stranger class were aillts and taeogs living in separate taeog trefs, though, according to the surveys, sharing, often in common, certain rights of grazing over certain districts with gwelys of tribesmen. Now these groups of taeogs and aillts were, according to the Codes, as we have seen, of two classes, and we recognise the same two classes when we find in the surveys not only groups of taeogs in taeog-trefs but also gwelys of non-tribesmen.

The normal group of the taeog-tref differed from the free tref in the fact that in it no family rights were recognised. All the members of it shared in its rights and payments equally per capita, and not per stirpes. They were all liable as a body, few or many, for the whole amount of the dues to the chieftains. During their fathers’ lifetime sons shared pari passu and equally with their parents, and other members of the group, in the pasture and common ploughing, except youngest sons, who remained with their fathers.

In the gwelys, on the other hand, as in the gwelys of tribesmen, there was recognition of family or blood relationships, and a patriarchal element.

There were thus under Cymric tribal custom various subordinate grades or classes. Beginning at the bottom of the ladder were:—

(1) The slaves who could be bought and sold, and who were reckoned as worth one pound of silver.

(2) The taeogs and aillts or permanent nativi, born non-tribesmen, without recognised family rights.

(3) Non-tribesmen growing or having grown in four generations into gwelys of non-tribesmen with recognised family rights.

(4) Strangers of exceptional position who, having married into the tribe, had become tribesmen in the fourth generation by repeated intermarriage.

And once more the fact should never be lost sight of, that the gradual growth into tribal or quasi-tribal rights was not a growth into exactly what in a modern sense would be called individual freedom. It was accompanied by the growth of ties which bound the family to the chieftain, till at the moment that at the fourth generation the recognition of rights of kindred was attained, the family found itself, as we have seen, so closely tied to the chieftain and the land that the newly recognised gwely had become adscriptus glebæ.

Finally, the tribal logic of the case was probably something like this:—

The free tribesman is the man who belongs to a kindred who can protect him by oath and by sword. Until a stranger has kinsmen who can do this he is an odd or kinless man, protected only by his lord. If he be killed his galanas goes to his lord; he has no recognised kin to receive it. If, on the other hand, he is charged with slaying another, he has no kin to swear to his innocence, the oath of a non-tribesman not being held good as against a tribesman. If guilty, he has no kin bound to fight in the feud for him, or to help him to pay a galanas for his crime. So that even when at the fourth generation the descendant of the alltud becomes the founder of a gwely he has gained only half the status of a tribesman. It is not till the fourth generation of descendants in the gwely, i.e. the seventh generation from the original settler, that a complete kindred has grown up. It is not till then that the descendant of the original alltud is surrounded by a full group of relatives, born in his great-grandfather’s gwely, whose oaths can be taken and who can protect him by oath and sword or in payment of galanas. All this time the alltud family have been more or less dependent on the protection of the chieftain, and rights and obligations are apt to be correlative.

The object of this essay is to inquire how far, in the case of other tribes, evidence may be found of the working of somewhat similar tribal instincts, resulting in customary rules more or less like those of the Cymry, so that at last, turning attention to the Anglo-Saxon laws, we may be able all the more fully to recognise and appreciate in them the traits of tribal custom, which among other factors went to the making of Anglo-Saxon England.

In the meantime, for future reference, the following list of the galanas of various classes will be found convenient:—

CHAPTER III.
THE EVIDENCE OF BEOWULF ON TRIBAL CUSTOM REGULATING FEUDS &c.

The object of the short study, in this chapter, of Beowulf, is to learn what incidental information it may give of tribal usage regarding the blood feud, especially on points which, in the case of the substituted wergeld, present doubt and difficulty.[59]

Allusion has already been made to some of these points. Did the rule excluding galanas or blood-fine within the kindred extend beyond the gwely to the greater kindred? What happened to a tribesman in a feud between his paternal and maternal kindreds? Did he abstain from taking sides, or did a marriage so far unite two families or kindreds as to make them one for the purpose of blood-fine or feud, so as to prevent the feud or blood-fine from arising?

These are questions upon which we want light from the point of view of Welsh tribal custom, and upon which we approach Beowulf for light, with eyes open also to other matters of tribal usage as they may turn up.

Beowulf for the present purpose may be taken as an Anglian or Northumbrian recension of a story founded upon Scandinavian tradition, and designed for use or recital at some 8th century royal court—possibly, if Professor Earle’s suggestion be correct, that of King Offa.

The western horizon of the story extends to the Frisian shores, but the scene seems chiefly to lie in the Baltic.

The plot involves tribal relations between a chieftain of the Danes possibly of Zealand, and two Swedish chieftains. The two latter concern us most, and they seem to be the chiefs of two kindreds—Geats and Swedes—Beowulf himself being the link between them, his mother having married from one into the other kindred. This marriage at any rate was one between two kindreds.

There is no apparent effort on the part of the poet to enlighten the reader or those who heard him either upon the pedigrees of the persons mentioned in his story or upon the rules of Scandinavian tribal custom. But it happens that, by incidental hints dropped in the telling of the tale, the pedigree of each of the kindreds involved can be fairly made out, and has already been made out by translators and critics.

And as the story involves a homicide within Beowulf’s maternal kindred, and fighting and bloodshed between the kindreds in spite of the marriage link, and as it deals also with outside feuds, it happens to present remarkable opportunities for studying the action of tribal custom in various cases.

The evidence it gives is made all the more valuable by its being an Anglian version of Scandinavian traditions, inasmuch as the poet, or his Anglian interpreter, assumes throughout that the laws of the game, under Scandinavian tribal custom, were too well known to need explanation to his Anglian audience. So that by inference it would seem that the customs of Baltic chieftains were familiar at the court of Offa, and not very far removed from those of Anglian tradition.