1. This essay, based on the analysis of a very large number of scattered and partly unpublished data, attempts only to show the relationships among various enigmatic ques tions and to indicate the perspectives which seem to emerge from these questions because they are so related.

2. Monumenta Germaniae SS. IV, 701-2.

3. Mansi, XIII, 370-71.

4. Monumenta Germaniae LL. I, 24-28.

5. Among the other texts in which allusions to these practices can be found, we have particularly the correspondence of Pope Zachary (Monumenta Germaniae, Hist., Epit., Vol. III), recalling various times when Christian priests were not permitted to sacrifice bulls, goats, etc., during funeral ceremonies. It is even specified that "Equi selvatici multo amplius evitandi sunt" in one of these letters, all written during the middle of the eighth century and consequently contemporaneous with the Council of Leptines. A hundred and fifty years earlier, Gregory the Great recalled, in a letter to Brunhild, the anathema pronounced against ceremonies held around an animal's head. And in the monuments de scribed later in this article, the importance of horse or ram heads will be seen clearly.

6. This evidence is corroborated by the often reported presence of votive fireplaces in the Frankish cemeteries of the Rhineland (Meckenheim, Andernach, Gohr), of Lorraine (Bouzonville), and of Belgium (Franchimont). But it must be remembered that these cemeteries are always rather late (eighth century). It is difficult to conceive, however, that such practices, so consonant with the protohistoric mentality, were born spontaneously at the end of the Merovingian era. Consequently they probably experienced only a recrudes cence at the time, whose beginnings are hinted at in the acts of the last council of Toledo. Indeed, from 589 to 653, the Spanish councils scarcely allude to anything except the survival of magic. On the other hand, they show, during the reign of Receswinthe (653- 72), an appreciable deterioration of the Catholic organization. The immediate conse quence of this was the rise of paganism witnessed by Father Valerius in the years 680-90 (Span-Forschungen der Görresgesellschaft, Ist ser., III, 439-49). The mass influx of lay pagani into the monasteries had the unexpected result of paganizing the monks (Migne, P.L., LXXXVII, 439, 444, 447).

7. For the earlier period there is no lack of evidence concerning Spain. Canon 69 of the Capitulary of St. Martin of Braga states: "Non liceat Christianis prandia ad defunctorum sepulchra deferre et sacrificia reddere mortuorum Deo." The custom to which this inter diction referred has left its mark in Spain on many Christian cemeteries of the fourth and fifth centuries. For example, semicircular banquet tables were found at Tarragona, set around tombs provided with a vertical chimney which allowed food to be brought into the sepulchral cavity. It should be noted that one of these tombs, although sealed, was empty. It was a cenotaph (Span-Forschungen der Görresgesellschaft, V, 74-88).

8. Cemeteries are rarely cleared for archeological reasons, but when they are we find in most cases the offering silos to which we call attention here. Moreover, they are well known to the local people who often refer to them by a regional word such as ponnes, ouilles, etc.

There is no over-all study of them. They have always become incidentally the subject of some description in the regional monographs or in articles on other subjects. Among these rather numerous publications, we should cite at least:

Mouret, Sulpice-Sévère à Primuliac ("Bulletin Société Archéologique Scientifique et Littéraire de Béziers," Vol. XXXVI [1906]).

Coutil and Baudoin (Actes du Congrès Prehistorique de France, 1912, p. 798). Martellière (Bulletin Sociêté Archeologique et Historique de l'Orlianais, XI, 382).

E. Socley (Bulletin Société Prehistorique de France, 1912, pp. 308 and 745).

Dusan et Lacaze, "Mottes du Sud-Ouest recouvrant des constructions" (Revue Arché ologique du Midi de la France, 1866-67).

Grellet-Balguerie, "Le Souterrain-refuge de Mazères-Fiac" (Revue Archéologique du Midi de la France, 1866-67).

9. One must take care not to confuse these offering silos with the real or so-called funeral wells which have been reported in various areas of the ancient Roman world.

10. Ledain, "L'église des Jacobins à Thouars" (Bulletin de la Société des Antiquités de l'Ouest," 1889, p. 496).

Mariano Ribas Beltràn, Las excavaciones de Matoró ("Reunion de la Comisaria Provin cial de Excavaciones Arqueológicos de Barcelona," 1957).

M. Broëns, "L'église et le site antique de Saint-Cizi" (unpublished), and L'église de Saint-Pré de Lestelle (St.-Gaudens, 1946); cf. above Dusan et Lacaze, Mottes du Sud-Ouest.

11. Protohistoric tumuli were often selected as cemetery sites, both in ancient times and in the Middle Ages, but even in such cases it is immediately clear on first inspection that we are dealing with works which are different from mounds.

12. The fullest documentation that can be had in France on the mounds discussed here has been collected by the Commission pour l'Étude des Camps et Enceintes, established at the beginning of the century within the Société Préhistorique de France. Unfortunately this repertory, published by departments, gathers together terraced works of all kinds and of all periods, usually without any order. In addition to this source a large number of monographic notices, more or less detailed, are scattered in the periodical publications of the various national or regional French learned societies. It seems clear, nevertheless, that a multitude of mounds, still extant today, may never have been reported. Thus we have been able to locate in Bas Quercy over thirty mounds not listed in the various repertories in which about 140 others are included. Since almost all of the mounds are hidden in wooded areas, this shows that we might find many more by undertaking systematic prospecting.

For Central Europe we have the numerous studies published by M. Much, I. Spöttl, von Reviczsky, Riehl, Wiedermann, Hoernes, Woldrich, Graf Würmbrand, Trapp, Kondelka, Demitrijkiewiez, etc., in the Mittheilungen des Anthrop-Gesellschaft in Wien, during the last quarter of the nineteenth century.

During the same period, the Beiträge zur Anthropologie und Urgeschichte Bayerns published articles on Bavarian mounds by J. Ranke, A. Thiersch, S. Hartmann, Würdinger, Ohlenschlager, and Sepp. Most of these studies also deal with the hypogea, underground works which we shall discuss later and with which the mounds we are studying are closely connected.

The observations noted above were made during the excavations (still unpublished) of the mound of Esmes (Commune of Montesquieu, in Tarn-et-Garonne), of that at Pélauze (L'Honor de Cos, Tarn-et-Garonne), and of the one at Couchines (Beauville, Lot-et-Garonne). They have been corroborated by test borings made in many similar works.

13. These ollae are always well fired, despite the rough appearance given them by the quartz drier added to the earth from which they are made. They were formed in two parts, joined near the top of the belly in a flattened fillet, decorated with gadroons made with the thumb. The lower part, with a round bottom and no base, was hand-molded, while the neck and lip were usually turned. The pitchers, less perfectly fired, are made of an extremely fine and porous pâte. The ornamentation, often not present, consists of nipples stuck onto the belly, or of stamped solar emblems; the latter are found in pottery of the mounds of Central Europe, but are on the bottom. In the north of France, the corres ponding pottery is related to the so-called Pingsdorf type, characterized by several streaks of reddish slip on an ochre or white pâte.

14. See Fig. 6 [6].

15. In the fill-in soil to which these mounds owe the upper part of their profile, it some times happens that Roman coins, fragments of tegulae, or even fragments of chipped flints are found, and the finder has concluded in favor of an ancient or prehistoric origin for the monument, instead of confining himself to the admission that the site had been occupied long before it was modified by the building of the mound.

16. As at Le Verdier (Commune of Montauban, Tarn-et-Garonne).

17. At Bétricourt, for example (Commune of Rouvroy, Somme) (Terninck, Congrès Archéologique de France, Session d'Arras, 1880, p. 158); at Verteillac-Coutures (Dordogne) (Hardy, Bulletin Société Archéologique du Périgord, XIII [1886], 447). Apparently we have here examples of the custom to which the letter of Gregory the Great to Brunhild, mentioned in n. 5, refers.

18. For example, l'AlIée de la Justice, at Presles (Seine-et-Oise).

19. Mémoires de la Société des Antiqttités du Centre, 1882, p. 2; report by Cartailhac in Matériaux, 1885, p. 228.

20. Cf. M. Broëns, Le souterrain-refuge de la Combe-Nègre ("Bulletin Archéologique du Comité des Travaux Scientifiques," 1938). The incorrect term souterrain-refuge ("cave used as refuge") had at that time been forced on the writer, who now formally rejects it. See also the writer's Un enigma arqueologico: los hypogeos de Cataluña y sus semejantes en el conjunto de Europa Occidental (Ampurias, 1960).

21. Souterrains-refuges-a better term would be refuges souterrains-with which these hypogea have generally been confused in France, are, however, distinguished from hypogea by nearly all their features. The best known are those of Artois, generally designated locally by the terms boves or muches. They are veritable underground villages or towns, made up of rows of rather spacious cells, set side by side along rectilinear corridors wide enough to permit several people to pass abreast or even for cattle to go through. They always have several exits and never show any of the enigmatic features, such as manholes, that we find in hypogea. Everything in them is logical and functional.

We know, also, that the ground under most medieval cities was crisscrossed in all directions by long galleries. Rather than serving as refuges, these galleries were probably used to evacuate the population in case of fire.

In the Lérida plain, which is absolutely barren and where all the population is widely scattered, most of the feudal centers were equipped in the twelfth century with subterranean passageways having secret exits in the middle of the countryside (Al Himyari, Kitab Ar-Rawd, published by Lévi-Provençal, La Péninsule Ibérique au Moyen Âge, p. 202 and n. 3). These galleries, thus clearly different from the hypogea, were also very numerous in Catalonia and served to evacuate personnel in case of surprise attack.

As for troglodytic dwellings, their chambers are often quite like those of hypogea, with side benches, niches, and the same careful execution of the work—but their arrangement is different. They are not built in depth but are placed at the level of a slope or cliff, in order to let in as much light as possible through their doors and windows.

The non-utilitarian character of hypogea was immediately apparent to the archeolo gists of Central Europe who dealt with these monuments in the last century. How did it escape most of the regional researchers or amateurs in France who explored some of them? No doubt because these archeologists had to limit their study to the fortuitous discoveries made in each one's own province. Moreover, since the publication of A. Blanchet's Souter rains-refuges de France (Paris, 1920), the edge of their curiosity seems to have become quite dulled. The learned numismatist stripped the question of all its mystery and, in a kind of offcial way, ratified the confusion between the underground works we are de scribing here and the refuges (cf. n. 42).

22. Bulletin Société Archéologique du Périgord, XIII (1886), 447.

23. Most of the hypogea of Périgord, like the one at Verteillac-Coutures, have a stair way bent several times at a right angle. But one, two, or three circular cells open onto this stairway, as at La Brugère (St. Michel de Villadeix, Dordogne), at Chalais, and at St. Pierre de Frugie (in the same department).

24. Most of these hypogea are under mounds, at Kissing, Lulling, Almering, Rocken stein, Mergentau, Julbach, Rottbach, Albersdorf, Nussberg, etc.

25. For exampe, at Dalhué (Gracay, Cher), at St Suipice-le-Donzeil (Creuse), and at Hautefaye (Issoudun, Creuse).

26. For example, at Roschitz I, II, III, V, VIII, at Watzendorf, at Dobersberg, and at Ober-Grünbach (Lower Austria). These circular galleries are strangely reminiscent of the cavity of the same form and dimensions of some of the Balearic megaliths called talayotes, like that of El Hostal.

27. This profile is particularly remarkable in the Department of Tarn-et-Garonne, at Espinas (Puygaillard), at Bosc-Grand (St. Bauzel), at Les Proats-Hauts (Montauban), and at St. Sernin-d'Ordalilles (St. Nauphary).

28. Observed particularly at Laversines (Oise), La Croux and Lestiol (Puycornet, Tarnet-Garonne), and La Bénèche (Caussade, Tarn-et-Garonne).

29. The "gratuitous" nature of the stairway is evident at Barraves (Caussade, Tarn-et-Garonne).

30. The morphological differences between hypogea of neighboring areas are some times such that one would not hesitate in seeing in them works of the same nature. In fact, from one type to another, the relationship is quite continuous, as is proved by the existence of hypogea which include, collectively, all the characteristics of each type. Con sequently we could not consider the most complex underground works as refuges, if the others are not. And moreover the history of the provinces confirms what the ethnologists have been able to declare in our own times, that the refuge on which dispersed popu lations most willingly rely during troubled times is the forest. In order to meet the danger of surprise, all that was needed was a dry, ventilated cave, well lined with masonry, with a secret entrance inside the house itself and an exit hidden in a ravine or in the woods. In any event, a person in flight who might have hidden in one of the hypogea we are dis cussing would have realized that he was in danger of being smoked out or simply walled in, however disconcerting the zigzags, corridors, manholes, and the quantity of chambers may have been. No aggressor, a priori, would have had the pointless imprudence to get involved in them.

31. Observations made at Esmes (Montesquieu) in particular, and at Pélauze (L'Honor de Cos, Tarn-et-Garonne).

32. Among the numerous published examples, the most convincing is the one observed by Grellet-Balguerie at Mazères-Fiac (Tarn) (Revue Archéologique du Midi de la France, 1867, p. 183).

33. At Combe-Nègre, the cavity prepared in this manner for the olla was in the left rear corner of each cell bench.

34. The age we have ascribed to this furniture corresponds rather closely to that which Caravan-Cachin in the Albi area and Pagès-Allary in Cantal gave it according to the stratigraphy of sumps, certain levels of which were well dated. Moreover, the chronology obtained in this manner is confirmed in a hundred ways, sometimes by structural rela tionships between hypogea and Romanesque churches, as in SS. Justo and Pastor (Barcelona City) and in San Cristo (Villasar, Barcelona Province); sometimes by ornamental details, as in the case of St. Sernin d'Ordalilles (St. Nauphary, Tarn-et-Garonne) where, in the corner of one chamber, a pillar of rectangular section was worked in the rock, with a base and crude capital. In addition, hypogea are particularly numerous in certain forest terrains, the toponymy of which clearly shows that the clearing of the land around monastery barns was not begun before the tenth century, as the documents prove. This observation is particularly apparent in Bas Quercy, in the old forests of Eysartens and of Moissac.

35. La Chesnaye (Revue des Traditions Populaires, XXX [1906], 170).

36. For example, in the hypogea of Barraves and of La Bénèche (Commune of Caussade, Tarn-et-Garonne).

37. After the introduction of Christianity into Germany, ruins there retained only a cryptographic and magic character.

38. These chimneys are also found in ancient third- and fourth-century tombs, both in Spain (see n. 7) and in Gaul, where the cemetery of Les Dunes, near Poitiers, has fur nished a number of examples (Mémoires de la Société des Antiquités de l'Ouest, 3d ser., XI). This whole question has been treated by W. Habery (Festschrift für August Oxé, 1938).

39. This was the case at the entrance to the last cell of Morthomiers.

40. Among the most curious in this respect are those of Almering bei Mühldorf, of Dünzelbach, and of Schwartzach (Bavaria), of Münzkirchen II and Mayrhof (Upper Austria), and of Csejthe (Slovakia).

41. Commune of Puygaillard (Tarn-et-Garonne).

42. For Danubian Europe, the bibliography on hypogea is the same one we listed in summary form in n. 12, in reference to mounds. It is necessary, however, to insist on the importance of the excellent book by P. Karner, Künstliche Höhlen aus Alter Zeit (Vienna, 1904), the only work which takes into account nothing but direct observations, which are all the more objective because the author refused to grant himself any competence in archaeology.

In France, the only work which has attempted to constitute a complete treatise is that of Adrien Blanchet, entitled Les Souterrains-refuges de France (Paris, 1920). As a simple compilation of all monographs previously published in France on caves of all kinds, this work is still useful because of the list it furnishes.

43. It is significant that "feudal" mounds have almost always retained the name of a lord, and that we almost always find them mentioned in medieval documents, whereas hypogea mounds are nameless today and apparently always have been.

44. P. Traeger, Verhandlung des Berliner Gesellschaft für Anthropologie Ethnographie und Urgeschichte, p. 52; see also the description of a hypogeum at Spata (Mesogea, Greece) in Bulletin de Correspondance Hellénique, I (1877), 261, and II (1878), 185.

45. The summary figuration of a house on the discoid steles which are found in cer tain Catalonian cemeteries of the eleventh century (at Tosa de Mombuy, particularly, and at Figuercla, Igualada district, Barcelona Province) perhaps took the place of the hypogeum which could not be dug in the hard rock substratum. The possibility is not ruled out that this expedient was resorted to in other cases, and even on a general scale, in certain regions where it seems to be clear that there never have been any hypogea.

46. E. Linckenheld, Les stèles funéraires en forme de maison chez les Médiomatriques et en Gaule ("Publication of the Faculté des Lettres de Strasbourg," Fasc. 38 [1927]).

47. Suburb of Mataró, Barcelona Province (M. Ribas Beltràn, El poblat iberic de Burriac i la seva necrópolis [Mataró, 1931]).

48. Cartailhac, Les âges préhistoriques de l'Espagne et du Portugal.

49. Gregory of Tours, In gloria confessorum, pars. 16 and 64.