^{page 15 note 1} Durkheim, E. and Mauss, M., Primitive Classification (Chicago, 1963).Google Scholar

^{page 15 note 2} Whorf, B., Language, Thought and Reality, ed. Carroll, J. (Cambridge, Mass., 1964).Google Scholar

^{page 15 note 3} Berlin, B. and Kay, P., Basic Color Terms, their Universality and Evolution (Berkeley and Los Angeles, 1969).Google Scholar

^{page 17 note 1} Mead, W., ‘Color in Old English Poetry’, PMLA 14 (1899), 169–206.CrossRefGoogle Scholar

^{page 17 note 2} Lerner, L., ‘Colour Words in Anglo-Saxon’, MLR 46 (1951), 246–9CrossRefGoogle Scholar repr. in Essential Articles for the Study of Old English Poetry, ed. Bessinger, J. Jr and Kahrl, S. (Hamden, Connecticut, 1968).Google Scholar

^{page 17 note 3} I use these terms not simply in their linguistic sense but also in the anthropological sense referring to abstract cognitive structures.

^{page 17 note 4} ‘Color in Old English Poetry’.

^{page 17 note 5} Gummere, F., On the Symbolic Use of the Colours Black and White in Germanic Tradition, Haverford College Studies 1 (Haverford, 1899).Google Scholar

^{page 18 note 1} Ardener, E., Social Anthropology and Language, Assoc. of Social Anthropologists Ser. 10 (London 1971).Google Scholar

^{page 18 note 2} Berlin and Kay, Basic Color Terms.

^{page 19 note 1} ‘Color in Old English Poetry’.

^{page 19 note 2} Willms, J., ‘Der Gebrauch der Farbenbezeichnungen in der Poesie Altenglands’, Inaugural Dissertation (Münster i. W., 1902)Google Scholar. Since this article was completed, I have come across König, G., ‘Die Bezeichnungen für Farbe, Glanz und Helligkeit im Altenglischen’, D. Phil, thesis (Mainz, 1957)Google Scholar. This offers by far the most comprehensive and accessible collection of data on the subject so far available. Dr König is very much concerned with collocation restrictions and makes many useful points, partly in agreement with my own views. We differ chiefly in the treatment of the ‘problem words’ (see below), where Dr König's atomistic approach is opposed to any attempt to set up wider correspondences between members of lexical sub-sets. It is my contention that only through the reconstruction of such sub-systems can we ever arrive at an understanding of the processes at work and at an appreciation of the logical consistency of the Old English colour system.

^{page 19 note 3} ‘Colour Words’.

^{page 20 note 1} E.g. Evans, R., An Introduction to Color (New York, 1948).Google Scholar

^{page 20 note 2} See Ardener, , Social Anthropology, esp. p. lxxvii.Google Scholar

^{page 20 note 3} If we wish to regard bar as limited to contexts marked ‘animate’, there is a problem in the BT Suppl. entry IIa, ‘the word occurs often as epithet of stones and trees used as boundary markers’. The other uses, however, can be safely dismissed as late or consciously poetical. For example, ‘of clife harum’ in The Metres of Boetbius(5113a) is not in the prose version, and we would have to be extremely careful if we drew any conclusions from use in the riddles, since, as I show elsewhere Barley, N., ‘Structural Aspects of the Anglo-Saxon Riddle’, Semiotica 10 (1974), (143–75)CrossRefGoogle Scholar. riddles manipulate just such restrictions to deceive or help the solver.

Are we, then, to regard the use of bar with stones and trees as an example of a word already partly freed of collocation restrictions, pointing the way for the other colour terms in their drift towards more general applicability? To adopt this view, it would be necessary to show that such occurrences are relatively late or colloquial. This may, indeed, be the case. The limitation of this phenomenon to boundary symbols suggests, however, another explanation and immediately warns the anthropologist that he is in danger of being misled. Boundary symbols, perhaps more than anything else, are likely to hold a special place in a people's world view. Modern works on classifications and taboos have noted the power of lines of demarcation; see, e.g., Douglas, M., Purity and Danger (London, 1966)CrossRefGoogle Scholar, and Van Gennep, A., Les Kites de Passage (Paris, 1909Google Scholar; trans, into English, London, 1960). Elsewhere (‘Anglo-Saxon Magico-Medicine’, Jnl of the Anthropological Soc. of Oxford 3 (1972), 67–77)Google Scholar I have shown that the Anglo-Saxon magico-medical system is based largely upon the manipulation and reaffirmation of boundaries. We see further concern with such matters in the concept of the outlaw, cast outside the social world and symbolically identified with the wolf. It is doubtless no coincidence that the charters that contain the bar stan phenomenon also indicate the presence of ‘wolfshead-trees’ on the liminal areas between properties. (An interestingly late case is of a dispute between the abbots of St Albans and Westminster in 1437, recorded in the Annales Monasterii S. Albani. The latter abbot complained that the former had destroyed a gallows erected by immemorial custom in a place called Nomannesland in order to destroy his boundary marker.) Worship of such plants and trees is attested in the laws (II Cnut 5, 1). This complex problem cannot be fully treated here, but perhaps enough has been said to show that this usage of bar may be explicable only by reference to other systems of belief that would remove certain stones and trees from the ‘non-animal’ category. Once more we see the impossibility of dealing with all colour words on a single level.

^{page 21 note 1} See Cameron, A., ‘The Old English Nouns of Colour: a Semantic Study’, unpub. B.Litt. thesis (Oxford, 1968)Google Scholar. I am grateful to Dr Bruce Mitchell of St Edmund Hall, Oxford, for bringing this thesis to my attention. For a note on færþu, see Szemerényi, O. ‘A New Leaf of the Gothic Bible’, Language 48 (1972), 1–10.CrossRefGoogle Scholar

^{page 21 note 2} Ploss, E., ‘Die Färberei in der germanischen Handwirtschaft’, Zeitscbrift für die Philologie 75 (1956), 1–21.Google Scholar

^{page 21 note 3} Berlin, and Kay, , Basic Color Terms, p. 37.Google Scholar

^{page 21 note 4} Ibid. p. 5.

^{page 21 note 5} It is regrettable that anthropologists have largely neglected investigation of this aspect of colour classification, since it provides an excellent means of access to other systems of classification operating within a culture.

^{page 21 note 6} Evans-Pritchard, E., The Nuer (Oxford, 1940).Google Scholar

^{page 22 note 1} A good study of such a specialized sub-system is Frake, C., ‘The Diagnosis of Disease among the Subanum of Mindanao’, Amer. Anthropologist 63 (1961), 113–32.CrossRefGoogle Scholar

^{page 22 note 2} Ullmann, S., The Principles of Semantics (Oxford, 1959).Google Scholar

^{page 22 note 3} For what follows I draw heavily on the evidence in E. Schwentner, ‘Eine sprachgeschichtliche Untersuchung über den Gebrauch und die Bedeutung der altgermanischen Farbenbezeichnungen’, Inaugural Dissertation (Göttingen, 1915), to which the reader is referred for more detail. This pioneering work brings together much of the available data on comparative Germanic colour studies. It is quite clear that Schwentner realized the existence of collocation restrictions on Germanic colour words but the absence of a structuralist approach prevented him from seeing these as anything but an arbitrary series of disjointed irregularities.

^{page 22 note 4} de Vries, J., Altnordisches etymologisches Wörterbuch (Leiden, 1961).Google Scholar

^{page 22 note 5} Schwyzer, E., ‘Germanisches und ungedeutetes in byzantinischen Pferdenamen’, ZDA 66 (1929). 93–9.Google Scholar

^{page 22 note 6} Unfortunately it would be misleading to use the evidence of Vulgar Latin forms as these are relatively late and the dangers of reciprocal borrowing and mutual influence are too great.

^{page 23 note 1} Schwentner, E., ‘Eine altgermanische Farbenbezeichnung’, Beiträge Zur Gescbichte der deutschen Sprache and Literalur 49 (1925), 423–9.Google Scholar

^{page 23 note 2} This raises the question of other colour sub-groups in Common Germanic times. I believe we can detect another group in *erpa, *blunda- and *haira-. Gmc. *erpa- occurs as ON jarpr, OHG erpf and OE earp. It remains fully active only in Old Norse, where it is almost totally restricted to human hair. Gmc. *blunda- is a vexed term that has, however, been traditionally accepted as a Germanic word borrowed early into Latin and referring to the hair of the typical Germanic warrior (see Ploss, ‘Die Färberei’). Gmc. *baira- occurs as ON barr, OHG her and OE bar. It is applied to human hair and animal fur. Although the evidence for these terms is less convincing than that for the horse words, it is at least enough to suggest that Common Germanic contained a group of colour words limited to human hair.

^{page 24 note 1} Dal, I., ‘German. brun als Epitheton von Waffen’, Norsk Tidskrift for Sprogvidenskap 9 (1938), 219–30.Google Scholar

^{page 24 note 2} Katz, J. and Fodor, J., ‘The Structure of a Semantic Theory’, Language 39 (1963), 170–210.CrossRefGoogle Scholar

^{page 25 note 1} This is why it has to be supplemented by an approach that is structural in the sense of Greimas, A., Semantique Structurale (Paris, 1966)Google Scholar. One does not, however, have to adopt as extreme a view of the differences between these two models as does Coseriu, E., ‘Zur Vorgeschichte der strukturellen Semantik: Heyses Analyse des Wortfeldes “Schall”’, in To Honor Roman Jakobson (Paris, 1967), esp. p. 493 nGoogle Scholar. Symbolic systems being of a multi-level order, we must expect to find nesting.

^{page 25 note 2} For a further example of this see Conklin, H., ‘Hanunoo Color Categories’, Southwestern Jnl of Anthropology 11 (1964), 339–44CrossRefGoogle Scholar. We should note a similar phenomenon in Modern English whereby wine is either ‘red’ or ‘white’ regardless of the fact that its actual colour may be purple, green or yellow.

^{page 25 note 3} ‘Die Färberei’.

^{page 26 note 1} Lest it should be thought that collocationally restricted sub-systems are unusual, let us note that Modern English again offers a parallel – besides the horse words already mentioned – in its (female) hair sub-set: blonde, brunette, auburn etc. The similarity of this to the Gmc *erpa-group is striking. It should be observed that the Anglo-Saxons were primarily concerned with the light–dark opposition even here, giving the system: fair = English = beautiful = freeborn, opposed to dark = Celtic = ugly = slave. In this way we can build up a picture of the way in which the Anglo-Saxons calqued a number of cognitive systems on a single axis of the colour system. I have suggested elsewhere (‘Anglo-Saxon Magico-Medicine’) that the hue axis may be used as a frame on which to hang a classificatory system of disease.

^{page 26 note 2} ‘The Old English Nouns of Colour’, introduction, passim.

^{page 28 note 1} Lévi-Strauss, C., La Pensée Sauvage (Paris, 1962)Google Scholar. Proof in such matters is impossible, but it seems not unlikely that the breaking down of collocation restrictions in the Old English period was connected with the great enrichment of material culture experienced by the Anglo-Saxons, firstly as a result of their invasion of these Romanized islands and secondly through their contacts with Mediterranean Christianity. Mauss, M. (Tie Gift (London, 1954))Google Scholar made the point that the fewer material possessions a people have the greater is the symbolic load carried by each. The converse also applies.