¹ Ventris, M. and Chadwick, J. (V.C.), Documents in Mycenaean Greek (Docs.) (Cambridge, 1956) 197 ff.Google Scholar

² For the identification of ideogram ^*145 as ‘wool’, see most recently Killen, J. T., ‘The Wool Ideogram in Linear B Texts’ in Hermathena xcvi (1962) 38 ff.Google Scholar; ‘Mycenaean po-ka: a Suggested Interpretation’ in PdP xvii (1962) 16 ff.

³ E. L. Bennett, J. Chadwick, M. Ventris (ed.), The Knossos Tablets: A Transliteration ¹(BICS Supplement No. 7, 1959). In the present paper, the system of transcription used in printing texts is that adopted in KT II. Readings are as in KT II, except for such changes as are included in Chadwick, J., Killen, J. T., ‘The Knossos Tablets: Corrigenda’ in Nestor (October 1st, 1961) 152 ff.Google Scholar, and in Killen, J. T., ‘New Joins of Linear B Tablets from Knossos’ in BICS ix (1962) 9 ff.Google Scholar The following corrections of KT II which are relevant to this paper are not listed in either of the above articles: Dg 926 .A: for pe. EWE read pe. RAM (over [[EWE]) (see n. 6 below); Dl 928 .B: for ki. EWE read ki. SHEEP; Dv 1153 .B: for o. EWE read o. RAM (see n. 7 below); D 7235: for o. EWE read se. EWE (this is clearly a Do text).

⁴ See Appendix A, below.

⁵ Docs. 401; Lejeune, M., Mémoires de philologie mycénienne, première série, 1955–7 (Paris, 1958) 73, 81, 129, 133.Google Scholar Cf. KN Ak 5767, where me. undoubtedly abbreviates me-zo ‘older’, me-wi-jo ‘younger’, or the like; also KN B 808, 814, 7039, where po. is likely to abbreviate po-ku-ta (cf. B 815, 816, &c).

⁶ As noted by Sundwall, J., ‘Zur Buchführung im Palast von Knossos’ in Societas Scientiarum Fennica, Commentat. Humanarum Litterarum xxii. 3 (1956) 4 f.Google Scholar KN Dg 926 .A is transcribed in KT II as: ] RAM 129 pe. EWE I; but the right-hand upright of the ‘EWE’ sign has apparently been erased, at least partially. The absence from the re maining upright of the two cross-strokes characteristic of male forms of livestock ideograms is no obstacle to belief that the scribe's intention here was to alter an erroneous Ewe to RAM; for although the absence of these cross-strokes may on occasion be significant (see n. 59 below), there are clear instances of these being carelessly omitted where the usual male form was clearly intended. Their omission here, therefore, in a context of (apparently) hasty correction, cannot be regarded as significant.

⁷ On Dg 1169 (an error ?). For o. EWE 2 on Dv 1153 .B (thus, KT II), read o. RAM 2.

⁸ V.C. note (Docs. 198) that at Pylos a total of 8,217 males, but only 1,554 females, are listed ‘on C-tablets sufficiently preserved’.

⁹ Nowadays ‘it is usual to allow one ram to every fifty or sixty ewes’ for breeding purposes: see Thomas, J. F. H., Sheep ³ (London, 1957) 57Google ScholarPubMed; and indications are that in medieval England a much higher number of breeding ewes was serviced by a single male: see Allison, K. J., ‘Flock Management in the Sixteenth and Seventeenth Centuries’ in Economic History Review xi. 1 (1958) 104.Google Scholar

¹⁰ Docs. 197 f. See also Finley, M. I., ‘The Mycenaean Tablets and Economic History’ in Economic History Review x. i (1957) 130Google Scholar: ‘In many, the excessive preponderance of rams over ewes and the frequency of round numbers in the totals show that these are not flock censuses’; Webster, T. B. L., From Mycenae to Homer (London, 1958) 19Google Scholar: ‘The occurrence of “deficit” both in Knossos and Pylos shows that many of these [livestock] tablets are records of tribute to the palace; a list in Ugarit gives the towns as supplying oxen as well as wheat and wine’; Chadwick, J., The Decipherment of Linear B (Cambridge, 1958) 122Google Scholar: ‘The sheep must be tribute, because a census is excluded by the round numbers and the calculated deficit.’

¹¹ Ibid. 195.

¹² See nn. 8, 9 above.

¹³ See Fraser, A. and Stamp, J. T., Sheep Husbandry and Diseases ³ (London, 1957) 153 f.Google Scholar

¹⁴ Ibid. 161 f. Flocks consisting entirely of rams were occasionally kept on medieval English manorial estates: see Allison, op. cit. 101.

¹⁵ See Thomas, op. cit. 65.

¹⁶ Cf. KN C 50.

¹⁷ ‘UDU.NITA est ce que nous appelons le “mouton”, de sexe masculin, mais sans insister sur ces capacités de reproducteur comme lorsqu'il s'agit du bélier, et même en le supposant coupé’: Bottéro, J., Archives royales de Mari VII (Paris, 1957) 248.Google Scholar Cf. Landsberger, B., ‘Studien zu den Urkunden aus der Zeit des Ninurta-tukul-Aššur’ in AfO x (1935) 155Google Scholar: ‘Im allgemeinen müssen wir aber annehmen, daβ, wenn männliche Tiere nicht durch einen Zusatz näher charakterisiert sind, kastrierte Exemplare gemeint sind.’

¹⁸ On Dw 931 + D 7293 (for this join, see Killen, op. cit. (n. 3 above) 10), 100 RAMS and 28 wool units are recorded, and no deficit in wool is listed. See also Dk 1613 + 5597 (n. 22 below).

¹⁹ But see Dk 5403 + 5562: .A ]RAM 98 WOOL 19 LB 2

.B ]RAM 2 WOOL 4[.

where although the o. which must have preceded RAM 2 in 1. B has not survived, the function of this second RAM entry is evident. On this text, see further n. 49 below.

²⁰ For a possible exception, see Da 1167+D 7194 (for this join, see Killen, op. cit. 11):

.A ] we-we-si-jo-jo RAM 157

.B ]-R0 di-ro o. RAM 1

For a certain exception, see Db 1606.

²¹ The numeral following ram on 1068 is read by the editors of KT II as 112; but although 112 would represent exactly four times the number of wool units recorded here (28), there is little doubt that 114 is the correct reading.

²² On Dk 1613 .A, wool 30 LB 1[ is presumably to be restored as wool 30 lb 2, which, allowing for the approximation to the nearest lb seen on Dk 1071, 1073, is one-quarter of 123. If a ram figure of 120 (or, less likely, 125) had formed the basis of calculation here, the wool total would presumably have appeared as 30 or 31 ( + ). For 1068, see above, n. 21: the wool total here has perhaps been miscalculated, since wool 28 lb i would have been a more precise approximation. It might just be conceivable that ram 115 formed the basis of calculation on this text, if sheep totals in anything other than multiples of ten were at all common at Knossos; but they are not.

²³ Over 800 of the texts listed in KT II have been as signed to the D class, and a large number of unclassifiable fragments are also likely to be parts of sheep tablets. Allowance must, of course, be made for the possibility that a number of texts now listed separately were originally parts of another tablet, also listed. But even when this allowance is made, any estimate of the total number of sheep recorded at Knossos cannot reasonably fall short of 80,000 (assuming that an average of 100 sheep was recorded on each tablet). And it may well be that a considerable part of the livestock archive has not survived. If, then, the sheep population of Mycenaean Crete was anything like the latest modern figure of 529,910 (see Primentas, N., ‘Greek Sheep and their Wool’ in Wool Knowledge iv. n (1959) 23Google Scholar: figures derived from the National Statistical Service of Greece Bulletin, 1956), it is extremely difficult to imagine that tribute of anything like 100,000 animals a year could have been extracted by the palace.

²⁴ Landsberger, op. cit. 153 ff., especially 154: ‘… die Kastration der Schafe in Babylonien bekannt, ja bei männlichen Tieren sogar die Regel war….’ Also, 155: ‘…der Grossteil der männlichen Tiere wird separiert und eigenen Hirten unterstellt.’ Cf. Oppenheim, L. and Hartman, L. F., ‘The Domestic Animals of Ancient Mesopotamia’ in JNES iv (1945) 156.Google Scholar

²⁵ Page, F. M., ‘Bidentes Hoylandie: A Mediaeval Sheep-Farm’ in Economic Journal, Economic History Series no. 4 (1929) 603 ff.Google Scholar; Power, E., The Wool Trade in Medieval English History (Oxford, 1941) 7Google Scholar; Smith, R. A. L., Canterbury Cathedral Priory (Cambridge, 1943) 148–55Google Scholar; SirClapham, J., A Concise Economic History of Britain (Cambridge, 1957) 106Google Scholar; Trow-Smith, R., A History of British Livestock Husbandry to 1700 (London, 1957) 148–50Google Scholar; Allison, op. cit. passim: Waites, B., ‘Wool Production on a Medieval Yorkshire Estate’ in Wool Knowledge iv. n (1959) 17Google Scholar; Simpson, A., The Wealth of the Gentry, 114.0–1660 (Cambridge, 1961) ch. 5.Google Scholar

²⁶ M. Griffith in Thomas, op. cit. 150.

²⁷ Fraser and Stamp, op. cit. 155, 298.

²⁸ Ibid. 198 f.

²⁹ Ibid. 155.

³⁰ Griffith notes (loc. cit.), with reference to Welsh mountain sheep, that ‘a strong wether would yield a fleece of 4 to 5 lb., whilst a ewe would only average about lb.’. Cf. Allison, op. cit. 105, and n. 58 below.

³¹ See below, pp. 9 ff.

³² Adapted from Waites (loc. cit.).

³³ Docs. 198.

³⁴ See (e.g.) Walter of Henley's Husbandry (ed. Lamond, E. (London, 1890)).Google Scholar For the considerable extent to which the application of the theories of such writers is reflected in actual manorial accounts, see Drew, J. S., ‘Manorial Accounts of St. Swithun's Priory, Winchester’ in English Historical Review lxii (Jan. 1947) 20 f.CrossRefGoogle Scholar (I am indebted to Mr. E. Miller, of St. John's College, Cambridge, for drawing my attention to this article.)

³⁵ See n. 25 above.

³⁶ See (e.g.) Anon., The Office of Seneschal, in Lamond, E. (ed.), Walter of Henley's Husbandry (London, 1890) 99Google Scholar: ‘And the keeper of the wethers ought to have in his keeping four hundred wethers if the pasture be large, or more, if it is narrow, fewer; the keeper of the ewes ought to have three hundred in large pasture; the keeper of the hogs two hundred.’

³⁷ See (e.g.) Drew, op. cit. 24, n. 2: ‘Multones. Restauretur falda ita quod sint in toto ccl’ (quoted from the accounts of St. Swithun's Priory, Winchester, and relating to its flock at Hannington in 1324).

³⁸ See (e.g.) ibid. 24: ‘For example, ifit is thought that a manor could support more cattle or more sheep, it is the auditors (and not the steward) who give instructions for additional stock to be bought or bred.’

³⁵ ‘They are the records of Thomas Townshend, who had been entrusted by his father, Sir Roger, with four flocks at the end of 1544; a wether flock at East Rainham, two breeding-flocks at Kipton and Barn (?), and a hogg flock at South Creake; amounting in all to 3,000 sheep. But he altered the whole layout in the first year and increased its total by a third. One of the ewe flocks was sold off to a Norfolk breeder, while the other, at Kipton, was doubled. The flock at East Rainham was doubled and another flock of the same composition (480 wethers, 120 wether hoggs, 120 rams and riggs) was started at Shereford. Only at South Creake was the Michaelmas figure of the previous year unaltered—1,200 hoggs and young wethers. The new total of 3,900 was maintained throughout 1546 and 1547 and then pushed up again in 1548 — to 4,200 – by more expansion at Kipton’ (Simpson, op. cit. (n. 25 above) 183; my italics). Other accounts show more random totals (see Simpson, op. cit. 190 ff., for records of the Culford flocks, 1558–1618). But since far greater numbers of sheep were being dealt with in the palace at Knossos, and since the mathematical ability of the palace officials seems to have been limited (see W. F. Wyatt, Jr., ‘The Ma Tablets from Pylos’ in AJA lxvi. 1 (1962) 41), the rigidity with which this system of flock management was apparently applied by the Mycenaeans is readily understandable.

⁴⁰ Allison, op. cit. 98; Drew, op. cit. 31, 33.

⁴¹ V.C., Docs. 401.

⁴² Primentas, op. cit. (n. 23 above) 20.

⁴³ See (e.g.) Fraser and Stamp, op. cit. 298: ‘Hill-sheep arms were divided according to local geography, waterheds, and the course of rivers into distinct “hirsels”, each hirsel numbering about 600 head of sheep under the charge of one shepherd.’

⁴⁴ I am indebted for information on this point to Dr. J. R. Goody of St. John's College, Cambridge.

⁴⁵ V.C., Docs. 201 f.

⁴⁶ Ibid. 57 f.

⁴⁷ Drew, op. cit. 30: ‘In 1299 and 1307 they (the audit board) went further, and, for the first time, they stated the amount of the wool target. This was 100 fleeces to the wey at Houghton, and 120 to the wey on eight other manors. [‘This is roughly an average of lb. per fleece at Houghton, and lb. elsewhere.’] The reason for the difference no doubt was that at Houghton all the sheep kept were wethers, the fleeces of which would naturally average more than those of mixed flocks.’ Cf. Simpson, op. cit. 181; 185:’… a fleece of lbs. was common’; Allison, op. cit. 104–6.

⁴⁸ Wiseman, D. J., The Alalakh Tablets (London, 1953) 100.Google Scholar

⁴⁹ The flock listed on Dw 931+D 7293 (see n. 18 above) has presumably exceeded its target; and the similar absence of a wool deficit on Dk 1613+5597 (n. 22 above) shows that here, too, the target has been reached. At first sight, the likeliest explanation of the RAM entry of 114 and 123 animals (and no RAM deficit) on Dk 1068, Dk 1613 would seem to be that the two shepherds in question held precisely 114 and 123 sheep, and that wool assessments were based on actual, rather than notional, flock strengths. But it is then curious that on Dl records wool totals should obviously be calculated on the number of ewes (and ki. RAMS) both held and ‘missing’ (see n. 62 below); and it is also difficult to reconcile the recording of a deficit on Dk 5403+5562 (see n. 19 above) with this explanation of 1068, 1613. M. Pope, however, has pointed out (see ‘The Cretulae and the Linear A Accounting System’ in ABSA lv (1960) 204 f.) that a number of Linear A tablets exhibit totals (like 95 an ) which suggest that, for some reason, totals of 100, 50, &c, were on occasion reduced by 5 per cent. So far there has been nothing to suggest a similar practice in Linear B accounting; but it is of interest to note that 114 and 123 are 95 per cent, of, respectively, 120 and 130 (allowing in the second instance for the usual rounding off to the nearest whole number). It is possible, then, that these are not, after all, actual flock totals, but notional figures on which the shepherds in question were assessed for wool contributions. The absence of a deficit on 1068, 1613, and the presence of o. on 5403 might then be due simply to the fact that the shepherds listed on the first two texts did have in their charge the number of sheep on which they were assessed, but that the same was not true of the shepherd named on 5403, who although assessed on 100 rams, only held 98. (If this explanation of the ram totals on 1068, 1613 is correct, our earlier argument that these figures are difficult to reconcile with a view of the D tablets as records of tribute is not invalidated; for there is no evidence in the whole of the Da–Dg texts of similar 5 per cent, reductions; and it is difficult to see why reductions should have been made, in certain cases, when sheep and wool were required, but not when sheep alone were demanded.)

⁵⁰ For TI-MI-ZA here, see Bennett, E. L. in Nestor, Oct. 1st, 1961, 151.Google Scholar

⁵¹ We may here contrast the case of the personal name wa-du-na-ro. On Dc 1118, a shepherd of this name is recorded in connexion with the place-name ku-ta-to, and 100 sheep; and the name recurs on Db 1242, but this time in connexion with 110 sheep at tu-ri-so. In this instance, as in a number of others, we can only conclude that wa-du-na-ro refers to a different person on each tablet. A further possible instance of two records referring to the same flock is revealed by the comparison of Dk 1077+5292:

(For the join of 7228 and 7244, see BICS xi (1962) 15. Since the preparation ofthat article, X 1496 has been seen by the writer from photographs to join with 7228.) Whether ]AZ-RA-RO on Dk 1070 (the reading is certain, pace KT ll) is to be identified with ]ZA-RA-RO on Dd 1429+5264+5327 is unclear; the sheep total on either tablet is the same (100), but the place-names differ (ku-ta-to on Dk 1070, pa-i-to on Dd 1429). On the alternation of da-mi-ni-jo and da-mi-ni-jo ku-ta-to illustrated by the comparison of Dk 1077+5292 with X 1492+D 7228 + D 7244, and of Dk 1076 with Df 1121, see J. T. Killen, ‘Some Adjuncts to the sheep Ideogram on Knossos Tablets’ in Eranos (forthcoming).

⁵² See (e.g.) V.C., Docs. 114.

⁵³ If the reading o. RAM 4 on Dw 5228 .B is correct, it would seem probable either that by the time of the compilation of the shearing record of this flock missing animals had been replaced and the flocks brought up to strength (see RAM 50 on Dk 1071), or that these four animals had been lost since the shearing season, when this flock had been complete. If the two groups of records were compiled at different times, the fact that the information they record is contained in separate documents (in contrast, it may be noted, to the Dl records) is readily understandable.

⁵⁴ For an earlier discussion of their function, see Chadwick, op. cit. 122: ‘We can only speculate on what became of the sheep thus contributed; the altar and the kitchen can hardly have accounted for them all, unless meat was eaten on a much greater scale than in classical Greece.’ See also Howe, T. P., ‘Linear B and Hesiod's Breadwinners’ in TAPA lxxxix (1958) 54 f.Google Scholar Howe assumes that the ‘Pylos tablets supply more impressive evidence (for meat consumption) than the discovery of animal bones “in generous quantities in all Mycenaean sites”, since an inventory of these so far records a total of over 12,500 livestock, and, as Ventris and Chadwick comment, “there is good reason to think that the figures recorded only a small percentage of the total flocks”.’ But the recording of large numbers of sheep does not prove per se that meat was ‘the real staple of the Mycenaeans’ (Howe, ibid.), or indeed that it was eaten at all. On this point see, for example, Power, op. cit. (n. 30 above) 20f.: ‘we breed our sheep for mutton in England today, but in the Middle Ages meat was merely a use to which sheep not good enough to keep for wool or for breeding could be put …’. See also Heer, F., The Medieval World (London, 1962) 27Google Scholar: ‘Arable land was also becoming scarcer, since some of it was being turned into pasture for sheep. There was nothing inherently objectionable about sheep. Rightly used, they were hard-working and serviceable animals, providing dung to manure the ground, milk for cheese, skin for parchment and fleece for wool. But side by side with the peasants’ sheep there were now the lord's sheep, no longer beasts of all trades, but devoted only to the production of wool' (my italics).

⁵⁵ Since the great majority of these L (CLOTH) tablets are apparently in the same scribal hand as the majority of WOMAN records (Series Ak, Ap), it is perhaps tempting to think of the women listed on Ak, Ap texts as wool-workers.

⁵⁶ In ancient Babylonia, for example, ‘bei Fleischschafen das Geschlecht irrelevant war’ (Landsberger, op. cit. (n. 29 above) 153. The carcass weight of ewes and wethers does not differ appreciably; the fleece weight does (see nn. 30, 47 above).

⁵⁷ A number of Dl texts (see Dl 790, 791, 916, D 7072, 7076, 7082, 7086, 7092) list WE in place of wool; and the normal proportion of EWES: ki. RAMS: WE on such texts seems to be 2:2: 1 (see 7086). For discussion of the meaning of this ideogram, see Docs. 196, 208. If we does represent wetalon or welelon, ‘yearling’, it is not entirely inconceivable that, for every ewe in certain flocks, 15 young animals were demanded by the palace every year: see (e.g.) n. 61 below. But it may be that the produce of some (outlying?) flocks was collected every second year.

⁵⁸ See nn. 30, 47 above, and cf. Drew, op. cit. (n. 34 above) 32: ‘At Littleton in 1324 the reeve [an official] was charged 43s. 1d. because his fleeces did not average 2 lb. per wether or ram, lb. per ewe, 1 lb. per hogaster, and lb. per lamb.’

⁵⁹ On Dl 463, the number of ki. animals is probably half the number of EWES, and on 1060 this is certainly so. But note that in either case it is the undifferentiated form of the ideogram, not RAM, which follows ki. Cf. also D 7079, D 7087, C 7088. The precise significance of the undifferentiated livestock ideograms remains obscure. The evidence of KN Ca 895, where po-ro, ‘colt’, is followed by an undifferentiated HORSE ideogram, suggests that some variety of young animal may be denoted by SHEEP (see Lee, D. J. N., ‘Mycenaean Anato/Anata’ in PdP xv (1960) 405 n. 12)Google Scholar; and since breeding animals produce approximately equal numbers of males and females, the evidence of 1060 might be regarded as suggesting a meaning of ‘young male’ for ki. SHEEP, as against ‘young animal of either sex’ for ki. RAM. But this is highly speculative.

⁶⁰ Drew, op. cit. 28.

⁶¹ Walter of Henley's Husbandry (n. 36 above) 75. Cf. Allison, op. cit. (n. 9 above) 103: ‘The lambing-rate in Norfolk during these centuries was always below one lamb per ewe per annum; it varied between 0·5 and 0·8 lambs per ewe, occasionally reaching 0·9 towards 1700 when some slight improvement is discernible. Present-day rates rarely fall below 1·0 and often rise above 2·0 lambs per ewe per annum. The constantly low rate in Norfolk during the sixteenth and seventeenth centuries was due partly to the breed of sheep: even in the eighteenth century the Norfolks produced only one lamb per ewe per annum.’

⁶² For an earlier suggestion that the Dl texts are records of breeding animals and their young, see Sundwall, op. cit. (n. 6 above) 13 f. It might be regarded as curious that on Dl texts the WOOL target is apparently calculated on the basis of the number of ewes kept plus the number of ki. RAMS expected to be produced, regardless of whether they were produced or not. But for a parallel situation see Walter of Henley's Husbandry (n. 36 above) 75: ‘… and if there be a cow that has not calved or a ewe which has not lambed in the year, let it be inquired whose fault this is, either the bailiff's or the provost's or the keeper's, for want of keeping or for want of a male, or if the provost could have changed it for another in time and did not, and, ifit be found to be any fault of these, let them be fully charged for the issue or its value.’ For the practical operation of this precept, see Drew, op. cit. 35: ‘A manor would have (we will say) ten cows, and only eight calves were born this year because two of the said cows were sterile. The auditors would very likely decide that, in the case of one cow the sterility might have been avoided. If such was their view, they altered the “eight” to “nine” and the “two” to “one” on the roll. They then charged the official, not only 18d. or 20d. for the hypothetical calf, but also 3s. 4d. for the 65 lb. of imaginary cheese made from the milk that might have materialized but which did not’ (my italics). We can only speculate on what sanctions were imposed in order to encourage Mycenaean shepherds to reach their production quotas, in the absence of currency: possibly they lost their rights to distributions of other commodities by the palace for failure to do so.

⁶³ (a) No mention is made on the Dl texts of the rams required to service the ewes which are listed there; but this can be paralleled in medieval English records (see (e.g.) Allison, op. cit. 104), and is in any event unsurprising, since these are primarily lambing records, and the few rams which would have been required to tup the ewes in each flock would have been run separately at all times except when required for breeding.

(b) The number of Dl records which still exist can be only a small fraction of the original total (as is also true of the extant Dk texts). This is presumably an accident of survival; although it is possible that at the time of the destruction of the palace the records of lambing and shearing for the current year had not been completed. Evans (Palace of Minos iv. 943 f.) considered it likely that the final conflagration took place in March (though Pendlebury, (Archaeology of Crete (London, 1939) 231Google Scholar) regarded late April or early May as a likelier guess); and lambing and shearing would, of course, have taken place in the spring or early summer.

⁶⁴ For the reading of Do 927, see Chadwick, J., Killen, J. T., ‘The Knossos Tablets: Corrigenda’ in Nestor (Oct. 1st, 1961) 154.Google Scholar For a suggestion that the adjuncts pe. and za. on Do tablets are abbreviations of pe-ru-si-nu-wo ‘last year's’ and ^*za-we-te-io vel sim. ‘this year's’ (cf. za-we-te on PY Ma 225), see Minutes of the London [Mycenaean] Seminar, Nov. 21st, 1962.

⁶⁵ The equation of pa. with pa-ra-jo (first suggested by Furumark, A., ‘Ägäische Texte in griechischer Sprache’ in Eranos lii (1954) 29Google Scholar, has been criticized by Sundwall, J., ‘Zur Buchführung im Palast von Knossos’ in Societas Scientiarum Fennica, Commentationes Humanarum Litterarum xxii. 3 (1956) 4 f.Google Scholar, on the grounds that the adjunct is always found in conjunction with the RAM ideogram. But if RAM can represent ‘sheep of either sex’, this difficulty disappears; for if ‘old’ animals were disposed of between census and shearing time, there would be no need for an indication of their sex, which is only relevant to a calculation of a flock's potential wool production. That pa-ra-jo represents is now as certain as may be, although the supposition that and are derived from a common root with an initial labio-velar has led some to doubt this (for the development of ^*k^wa to Mycenaean qa, not pa, see Vilborg, E., A Tentative Grammar of Mycenaean Greek (Göteborg, 1960) 46).Google Scholar J. Chadwick has shown, however, by comparison of PY Sa 787: to-sa pa-ra-ja we-je-ke-a ₂ WHEEL ZE 30 MO I with PY Sa 843+1190+1270 (see AJA lxiii. 2 (1959) 131) to-sa we-je-ke-a₂ ne-wa WHEEL+TE ZE 20, that pa-ra-ja and J. T. Killen ne-wa are alternants in formulas; and as ne-wa certainly represents pa-ra-ja is unlikely not to represent On this point, see most recently Hcubeck, A. in Gnomon xxxii (1960) 668 f.Google Scholar For the practice of'casting’ or disposing of old animals from standing herds, often for slaughter, see (e.g.) Walter of Henley's Husbandry (II. 36 above) 97: ‘The bailiff ought, after St. John's Day, to cause all the old and feeble oxen with bad teeth to be drafted out ….’ ‘And again, let all the old and weak [sheep] be drafted out before Lammas …’ (my italics). And cf. Simpson, op. cit. (n. 25 above) 195: ‘30 crone ewes sold … £6. 0. 0’; and ibid. 185: ‘The stock—which was kept much longer than would be thought efficient in the eighteenth century—was culled for sales at various seasons, with lambs, hoggs, young wethers, fat wethers and crones (the old animals) being the usual categories ….’

⁶⁶ M. Doria, in Interpretazioni di testi Micenei ii (Trieste, 1958), translating a-ko-ra, a-ke-re, terms which appear on the Pylos SHEEP records, as ‘flock’ and ‘herds’ (rather than, with V.C., as ‘collection’, ‘collects’), and seeing in the Cn texts evidence for seasonal movement of herds from moun tain to valley, regards these as records of real flocks, not of tribute. Since the Pylos texts, like their Knossos counter parts, record a great preponderance of RAMS, and, on most texts, round numbers of animals, the function of each group is likely to have been similar; and the discovery that the Knossos D records are of real flocks must accordingly provide strong support for Doria's view of the Cn records.

⁶⁷ Kantor, H. J., ‘The Aegean and the Orient in the Second Millennium b.c.’, in AJA li. I (1947) 58 f.Google Scholar Cf. Buck, R. J., ‘The Minoan Thalassocracy Re-examined’ in Historia xi. 2 (1962) 132.Google Scholar The Greek mainland may have been the destination of at least some of the wool exported from Crete. Certainly sheep are far fewer on the Pylos than on the Knossos tablets, and although a group of texts from Mycenae (Series Oe) clearly deals with wool, textile records at Pylos are almost non-existent. If Knossos fell c. 1400 B.C., the mainland may not have been able to draw on Crete for supplies of wool after that date; and this may partially explain the great increase of Mycenaean trade with Troy after 1400, for Troy is also likely to have derived much of its wealth from textile production (see Biegen, C. W., Boulter, C. G., Caskey, J. L., Rawson, M., Troy (Princeton, 1950–) iii. 32 f.Google Scholar; iv. 19; Page, D. L., History and the Homeric Iliad (Berkeley and Los Angeles, 1959) 69 f.).Google Scholar Perhaps, too, the legend of Jason's voyage in search of the Golden Fleece echoes mainland attempts to discover new sources of wool in the Black Sea area after 1400.

⁶⁸ For a suggestion that olive oil may have been exported, see most recently Hutchinson, R. W., Prehistoric Crete (Pelican Books, 1962) 245.Google Scholar For other possible exports, see Buck, loc. cit.

⁶⁹ I am deeply indebted to Dr. John Chadwick, both for suggesting the Knossos D tablets as a topic for study, and for stimulating criticism and unfailing encouragement throughout the preparation of this paper. I am also most grateful to Dr. M. I. Finley for his comments on an earlier draft of the paper, and to Mr. E. Miller, of St. John's College, Cambridge, for information on animal husbandry in medieval England.