Production in the Rest of the Kyklades

For the rest, Delos must have looked first to its other Kykladic neighbors. We know that these islands produced grain, sometimes in quantity. Modern data certainly suggest their potential productivity. For example, in three consecutive years, between 1936 and 1938, the Kyklades produced 9,608,800, 17,639,100, and 16,575,500 kg of grain, equivalent (assuming the crop was virtually all barley) to roughly 186,000, 341,750, and 321,100 medimnoi.^[51] Early modern data collected by B. J. Slot point in the same direction. In 1670, Naxos produced the equivalent of about 8,200 medimnoi of grain, Paros about 19,000, Thera about 14,350, Melos about 10,680, Andros about 10,000, Syros about 4,900, or a total for just these islands of 67,130 medimnoi. From 1700 to 1718, an average of 42,813 pounds of wheat arrived from "the Archipelago" in Marseille; "the Archipelago" embraced more than just the Kyklades, but the figure gives a general sense of the availability of surpluses in the islands. This figure is the equivalent of about 480 medimnoi. Visiting the Kyklades at about the same time, J. Pitton de Tournefort reported exports in grain from los, Sikinos, and Naxos. Between the 1830s and 1911, Keos produced an average of 1,250,780 kg of barley, of which 110,000 kg were exported in 1906, equivalent to 24,230 medimnoi produced and 2,130 exported.^[52] These figures have no statistical value, but they suggest a small, but genuine, exportable surplus. Recent travelers and researchers continue to report the production of barley and wheat on the islands.^[53]

[51] S EE 1937: 112; 1938: 112; 1939: 112.

[52] B. J. Slot, Archipelagus turbatus (Istanbul, 1982), 1.302–6, 321–24. J. Pitton de Tournefort, A Voyage into the Levant (London, 1741), 1.266, 270, 228. Sutton in Landscape Archaeology, 390. Cf. however Sallares, 316.

[53] See Connell, In the Bee-Loud Glade, 18, 22–25; Wagstaff and Augustson in Island Polity, 110–12; Wagstaff, Augustson, and Gamble in ibid., 177–79; Wagstaff and Gamble in ibid., 98, 101; Halstead and Jones, JHS 109 (1989): 41–55; Sanders, BSA 79 (1984): 251–62; R. N. L. Barber, The Cyclades in the Bronze Age (Iowa City, 1987), 7. On soil types on Naxos and Paros, see J.-Y. Empereur and M. Picon, BCH 110 (1986): 496–98.

For an estimate of the potential productivity of the Kyklades, however, it is necessary to consider the factors that governed productivity over the long term. One of these, yield per hectare, has already been discussed, and I shall continue to use the figures in table 4.3. The other two are rainfall and area cropped. Although already discussed, the effect of insufficient rainfall on crops must be considered in more detail here because (1) rainfall is sufficient for barley production most of the time in the Kyklades, and (2) we can trace variability not only among years but also among islands, which implies that shortages on one island because of insufficient rain need not have meant shortages elsewhere. Area cropped is crucial, because the Kyklades today are apparently grossly underexploited compared to antiquity; the potential for grain production is certainly considerably greater than the rather neglected present appearance of many of the islands would suggest.

Rainfall. Rainfall is important not only because the amount of rainfall determines the success or failure of a crop, but also because it gauges the relation of social expectations of consumption (discussed above) to actual production. If a region produces more grain than social constraints demand, there will be a genuine surplus: that is to say, a surplus that can be regularly absorbed by a local nonfarming urban population or exported.

For 1951–70, Naxos received a mean rainfall during the growing season (December–May) of only 208.5 mm, enough rain to support a wheat crop only 44 percent of the time (seven years out of the sixteen for which data are available). Barley fared much better: in only two years of sixteen (12.5 percent) did rainfall fall outright below 200 mm, and in two (or possibly three) others (12.5 percent [18.75 percent]) was rainfall marginal for barley culture (200–240 mm). A culture dependent primarily on barley would therefore typically have experienced stress every third year (table 4.4).

Thera probably experiences similar conditions, although lack of data makes inference dangerous: for five years for which we have seasonal data (1933–37), there was always too little rain for wheat, and too little for barley in three years; but this could be the same kind of pattern that prevailed on Naxos in 1972–77.^[54] Some data for eleven earlier years (1896–1907) illustrate extreme small-scale variability of precipitation. Rainfall at the town of Phira (= Thera) fell short of the minimum for wheat in eight years out of eleven, while it sufficed for barley in all but three years (27 percent). The pattern is thus virtually identical to that on Naxos. At Gonia on Thera, in contrast, rain was more than sufficient for both crops in all five years for which data were recorded (1901–5), while on the slopes of

[54] S EE 1937: 17–18; 1938: 17–18; 1939: 17–18.

Profitis Elias in the same period, the wheat crop would have failed three times (60 percent) and the barley crop once (20 percent).^[55] Phira sits on the western ridge of the island, Gonia lies about 4.5 km inland to the east, and Profitis Elias rises to 564 m about 1.75 km SSE of Gonia. Within a range of barely six kilometers, therefore, rainfall varied enormously (table 4.5). Years that saw crop failures at Phira brought quite enough rain to support farmers near Gonia. This local variability is crucial for understanding the extremely localized stresses that ancient farmers underwent and the frequent reports in our sources of apparently quite localized shortages or failures. Disaster at Phira (as in 1901) did not necessarily spell disaster a few kilometers away.

[55] P. Wilski, Klimatologische Beobachtungen aus Thera (Berlin, 1902), 89; partially reproduced in table 4.5.

These results are important. They suggest that the Kyklades could have supported a reliable barley culture, although wheat would typically have been more problematic. Considerable variability from island to island, and even from place to place on a single island, is also to be expected. This confirms the picture early travelers paint. Tournefort's detailed discussion of the condition of individual islands shows great variability among them in their productivity, but with a few exceptions—los and Sikinos—they had far more success with barley than wheat. Some, including los, Sikinos, and Naxos, are specifically said to have exported grain.^[56] These results are also very much in harmony with the recent emphasis among scholars of ancient Greek agriculture on the importance for ancient farmers of exploiting micro variations in weather, soil conditions, and other variables by holding small plots in different locations. Theran farmers who held land only at Phira would certainly have had cause to rue it early in this century; there is no reason to suppose conditions were any different elsewhere in the islands, or indeed in antiquity.

Another important implication of the rainfall data relates to the extensiveness of crop failures because of lack of precipitation. Given not only the great interannual variability, but also the potential for extremely localized variability, great general failures are likely to have been rare. Each year some communities on some islands must have faced shortages, but typically they could expect to make them up from the surplus of luckier neighbors. Long experience would have built this factor into the calculus of production. Storage and the short-range movement of grain were the mechanisms that helped the islanders cope; no doubt, too, they provided a large part of the status and wealth that the island elite enjoyed.^[57]

Area Cultivated. In the 1930s, typically only about 6–8 percent of the total surface area of the Kyklades was under cultivation in cereals.^[58] If comparable areas were cropped in antiquity, the islands' production could have satisfied only a very small population, approaching a high percentage of

[56] Tournefort, Voyage, 1.192, Seriphos produces only a small barley crop; 1.228, Naxos exports barley; 1.266, los exports wheat, likewise Sikinos (1.270); 1.273, Pholegandros is self-sufficient in grain, with a population of 120 families (perhaps about 600–720 persons); 1.284, Thera grows mostly barley (not surprising, given modern rainfall statistics); II.8, Kythnos grows lots of barley but little wheat; II.16, Keos "abound[s]" in barley; II.32, Andros has plenty of barley but imports wheat from Volos; II.41, Tenos again has plenty of barley but little wheat.

[57] Cf. Gallant, 143–96.

[58] S EE 1937: 112; 1938: 112; 1939: 112: in 1936, 14,401 ha (6.4 percent); in 1937, 17, 198 ha (7.6 percent); and in 1938, 18,581 ha (8.2 percent).

Delos's estimated population (see table 4.6). Three considerations, however, suggest that considerably more of the islands' land was under cultivation in antiquity.

Recent study of the Melian countryside has found at least 17 percent, and up to 58 percent, of land suitable for cultivation on the criterion of surface slope, although only 13.95 percent of the island was cropped in 1971.^[59] Anything from 18 to 75 percent of potentially arable land was therefore idle. It is reasonable to suppose that much of this land would have been brought under cultivation in antiquity. Furthermore, this area was land arable as it was, without improvement. As any visitor to the Kyklades knows, even the steepest slopes have been rendered suitable for crops by terracing. Unfortunately, terrace walls are notoriously difficult to date, and study of the landscape archaeology of the islands remains in its infancy.^[60] However, M. Brunet's careful examination of terracing on Delos suggests that the system there was in place by the fifth or fourth century B.C. The Delian system brings the arable surface of Delos up to about 250 ha, nearly 70 percent of the total area.^[61] If the Delian system was created in response to the same needs as the terraces of its neighbors, then it would be safe to assume that the Kykladic system as a whole was in place by the early Hellenistic age at the latest. Many terraces are idle today, largely because of emigration in the face of declining opportunities for local employment and inability of local farmers to compete with imported foods, but if brought into use, they would considerably raise the total potential arable surface in the islands. The existence of the terraces thus provides a strong argument for much higher ancient rates of cropping than attested today.

In 329/8 B.C. , Skyros, which covers 20,900 ha, produced 9,600 medimnoi of wheat and 28,800 of barley; Lemnos, with a total area of 47,600 ha, produced 56,750 medimnoi of wheat and 248,525 of barley. If only 8 percent of the surface of these islands were under cultivation, Skyros would have enjoyed yields of roughly 1,122 kg/ha and Lemnos of 3,960 kg/ha!

[59] Wagstaff and Gamble in Island Polity, 101; Malcolm Wagstaff and Siv Augustson in ibid., 106–7.

[61] Brunet, 149 with n. 27.

Clearly such figures are impossible. Yields per ha become reasonable only by assuming 20–50 percent of the total surface was cropped in grains. This result also strongly reinforces the view that considerably more of the islands' land was planted with grain in antiquity than modern figures suggest.^[62]

Table 4.6 presents estimates for the productivity of the Kyklades, Krete, and Greece as a whole, based on these considerations. For total area of the islands, I use the figure for the modern nomos of about 2,251 km² , or 225,100 ha.^[63]

At a conservative cropping of 20–40 percent of total surface, Kykladic production could have supported very roughly 50,500 to 100,000 persons. It must be born in mind that these figures represent, not maxima, but a population that the mean production of the archipelago could support over the long haul. Consumption estimates have already taken into account social storage of surplus food, so that in good years there would be an excess

[62] Contra, Eberhard Ruschenbusch, ANSP 13 (1983) 172, 174, who offers no argument for his pessimistic views. For an unpublished Athenian inscription extracting taxes in grain from (probably) Imbros, Skyros, and Lemnos in 374/3 B.C. , see Isager-Skydsgaard, 140.

[63] S EE 1981: 2–3, table 1.3. Exact figures differ in different sourcebooks; I have used the most recent official Greek figures readily available to me.

available for appropriation—whether by the elite, the government, the military, or outside powers—and in poor years imports would be necessary. Local shortages would be made up from local sources whenever possible, as at Phira and Gonia on Thera.

For a population of 2,600–9,100 persons, Delian demand represents 18.2 percent of the total typical supply on pessimistic assumptions, only 2.5 percent on optimistic, 7–8 percent on the mean. Could the Kyklades have provided this much grain?

The population of the Kyklades in the Hellenistic age can be estimated roughly. Ruschenbusch calculates a total population of 22,092 for fifteen Kykladic islands.^[64] This figure may be low. It reckons the ratio of adult males to the full population at 1:4, which is probably too pessimistic; if we figure instead 1:5, we get 27,615 persons. Ruschenbusch regards slaves as an unimportant component of island populations, but even if only the upper 7 percent or so of the population could afford them,^[65] that still implies 1,933 slaveholders in the islands; if each held on average just two slaves, they would have added almost 3,870 persons to the population. There must also have been some permanent metic population. During the great siege, the Rhodians counted 1,000 able-bodied adult male metics in contrast to a citizen population of 6,000 (Diod. 20.84.2). The Kyklades, less prosperous, must have attracted proportionally fewer resident foreigners: let us say no more than 10 percent of the citizen population, including wives and children; this would add another 2,750 persons. Finally, there would have been a steady population of transients: merchants, travelers, theoroi, ambassadors, soldiers, dramatic artists—perhaps another 10 percent. These adjustments yield a total population of about 37,000.

It should be emphasized that there are many uncertainties in these estimates. Beside the technical objections, corrections, and adjustments that may be applied to Ruschenbusch's figures and methodology, the larger structural issues Sallares has raised, to which I have already alluded, come into play. The two important Kykladic surveys that have been published (for Melos and part of Keos) agree that population reached its height in the fourth century and had suffered a retreat by the late second and first. On Keos this shrinkage can be traced also in the polis center of Koresia, which may have first begun to suffer during the Khremonidean War.^[66] Unfortu-

[64] E. Ruschenbusch, ZPE 59 (1985): 258–59.

[65] Ruschenbusch in Aux Origines de l'Hellénisme, 265–69; ANSP 13 (1983): 176–79, reckoning slaves at 1.5–3 percent of the total population.

[66] Sallares, 62–73. Malcolm Wagstaff and John F. Cherry in Island Polity 145–46, 252–53; T. M. Whitelaw and J. L. Davis in Landscape Archaeology 265–81.

nately, the third century remains generally obscure. In general, there are two possibilities: either population stayed high throughout the years of this study, so that whatever figures are accepted should be regarded as rough maxima; or population began to decline during our period, which would imply either a rising surplus^[67] (and so relief for Delos, dependent as it was on appropriating such surpluses) or, assuming that production declined in tandem with the population, a roughly unchanging situation.

Given these caveats, two inferences are clear. First, the Kyklades must normally have been more or less self-sufficient in grain. This should come as no surprise. Most Greek poleis were self-sufficient; it is the rare cities chronically dependent on imports, like Athens, that are the exception. Even if we have underestimated the Kykladic population by half, a doubled total of 74,000 still falls within the high end of the range of estimated carrying capacity.^[68] In general, ordinary Delian demand could be satisfied out of the marginal surpluses of its Kykladic neighbors. Second, even at the lowest levels estimated, Delos's population made it unique in the Kyklades. Its role in the islands as a religious center and an economic focus lent it special status, which in turn probably helped to attract the goods its inhabitants needed from its Kykladic neighbors. The most important implication of these calculations, however, is that Delos normally depended only on its nearest neighbors for its supplies of grain. The island had no need regularly to seek supplies from distant producers like the states of the Black Sea, or even Egypt; we shall see below that there is in fact virtually no evidence to suggest any dependence on the country of its Ptolemaic overlords for grain.