[207] Proportionality was the organizing factor of the water distribution systems of medieval eastern Spain. Each irrigator received water in proportion to the amount of land he held. The water to which he was entitled, moreover, was not a fixed amount per unit of land, but rather a proportional one which varied with the volume of the river. All irrigators shared in times of abundance and were equally deprived in times of drought. Likewise, the total debit of the river or spring was divided among the principal canals taking water from it, in proportion to the amount of land served by each canal. The water of each main canal was, in turn, subdivided proportionally among its branches. The proportional quotas for each carrier were expressed in an abstract measurement unit, usually the "thread" of water (fila in Catalan, hila or hilo in Castilian). A fila thus represented one share of the total amount of water in a given river (or sector of a river), spring, or canal.

The abstract unit fila was translated into the reality of volumes of water by the dimensions of physical structures -- diversion dams, irrigation canals, and divisors -- which divided the water that flowed in any channel proportionally among the branches into which that channel divided. In situations during which water was so short that these physical structures could not at the same time preserve the principle of proportionality and deliver water efficiently, equality was preserved by the institution of a system of turns called tanda, dula, and so on, whereby the irrigator, instead of drawing water at will, was obliged to irrigate [208] in a specific order and could not irrigate again until his next turn.

One of the advantages of the proportional system was that equitable distribution could be assured without measure of time or of delivery orifice. When water was abundant the farmer took it at pleasure. When water was scarce a turn was imposed, and as the quantity of water available grew less, the interval between successive turns became greater. But no measure was needed in this eventuality, either. As long as the irrigator did not waste water (and there were penalties to ensure that he did not), he could only use so much; he had only a certain amount of land and could not overwater his crops without harming them. Therefore the only control needed to ensure equality of partition was to see that he could not water again until everyone else had had a turn.

Of the three components of the system of proportional distribution -- the fila, the dam and divisors, and the turn -- only the last two were palpable. Divisors and turns assured, or were designed to assure, that water from the river would reach the individual farmer in an amount proportional to the area of land which he had the right to irrigate. These complex arrangements (antedating the Reconquest) were preserved through custom and by ensuring that the dimensions of divisors and canals were not altered. But the fact that rights were often expressed in filas, the abstract and imaginary unit, has caused no end of confusion among students of Spanish irrigation.

In traditional irrigation systems that function on the basis of custom, and where proportionality is the basis of water distribution, irrigators may not always understand the rationale of the apportionment. E. R. Leach's observations of such a system of distribution in modern Ceylon -- whose irrigation customs, like those of Spain, have developed over a period of more than one thousand years -- are relevant to the Valencian system:

It must be pointed out that from the Vel Vidâne's [irrigation officer] point of view the ratios [of distribution] are established by tradition; they are not something which he has to work out for himself. Moreover the ratios are not [209] thought of as fractions but as karahankota [wooden divisors] of different sizes with the grooves measured in inches. . . .

The complexity of the arrangement is itself relevant to my theme since such a system is virtually unalterable. Although the present generation of Pul Eliya villagers are not at all clear about the inner logic of it all they are keenly aware that the numerical formulae handed down from ancient times are very important. The general view seems to be: "We don't understand why things are arranged like this, but this is how they are, and we had better leave them alone." ⁽¹⁾

My analysis of the origin of the fila proceeds from one initial observation: the traditional quotas of filas of river water for different Spanish rivers, irrigation districts, or principal canals are typically expressed in multiples of twelve. Thus the Mijares, at the point where it serves the four towns of Castellón, Burriana, Villarreal, and Almassora, is said to carry sixty filas. ⁽²⁾ The river Sans at Játiva has twenty-four filas, ⁽³⁾ as has the river of Lorca. ⁽⁴⁾ The Royal Canal of Alcoy (serving Gandia and Oliva) takes forty-eight filas from the Serpis at the diversion dam of Alcoy. ⁽⁵⁾ The Moncada Canal of the Valencian huerta also carries forty-eight filas. These figures suggest that one fila is the equivalent of one hour of water, an assertion supported by the Marqués de Molin's definition of the hilo of Elche as "the twelfth part of the water which flows through the canal in the period of twelve hours." ⁽⁶⁾

[210] In cases where water is divided by time units as well as by filas, the equivalence of the hour and the fila is clear. There is such a twofold definition in the division of the water of the Mijares in 1346 (Table 18) by Prince Peter of Aragon, who de-

Assuming that the Mijares fila was "doubled" in times of drought (the dobla, or fila dobla, meant that the water ran for twice as long as usual), the equivalence of filas and hours is clear: one fila would equal two hours. Thus in times of severe water shortage the four Mijares towns instituted a turn in which each received water in the same proportion as its basic right (expressed in filas). In times of abundant water all four could draw water from the river concurrently, but the proportions would be approximately the same because the capacity of each canal would determine the share each received. The statutory proportions were, of course, the basis of the size of the canals, reflecting the amount of land irrigated. Burriana, the farthest downstream [211] of the four towns, irrigated more land than the others and therefore drew more water.

The fila, then, is a figure defining that proportion of the total debit which each canal takes (1) in times of abundance, according to its capacity, and (2) in times of drought, according to a commensurate number of hours. The traditional quotas of filas of the canals of the Valencian huerta present both a confirmation and a puzzle. Moncada and the seven lower canals received water in the amounts indicated in Table 19. The division of the lower canals into fourteens and tens suggests that these were originally paired, Moncada having two days' water then two lower canals having the next day, two the next, and so forth. If a fila was equivalent to an hour, then the traditional apportionment represented a turn of six days (or one week, if there was no irrigation on the Sabbath ⁽⁸⁾). There still remains the problem of identifying the huerta's actual allotment of a hundred thirty-eight filas with an expected six-day total of one hundred forty-four hours. The discrepancy is resolved by completing the pairs: in order to preserve the logically indicated order, there must once have been another canal on the left bank of the Guadalaviar between Mestalla and Rascanya and paired with Favara, with an allotment of ten filas, as in Table 20.

There is some doubt as to the correct allotment of Rovella. Most of the sources (following Jaubert de Passa) say fourteen, but José Martínez Aloy states that the correct figure is ten. ⁽⁹⁾ Ten filas is what one would expect for Rovella, according to the logic of the pairing. If fourteen is the correct figure, however, then one must assume that Rovella's original ten filas were increased at such a time when the lost canal between Mestalla and Rascanya ceased to function autonomously. The main canals of the huerta, running parallel to each other along the river, often serve the same area and even pass over and under one another. These canal networks can merge and separate with considerable ease: for example, when the diversion dam of the Rovella Canal was destroyed in the mid-eighteenth century, Rovella's service area was supplied through the Favara system, which diverted Rovella's filas as well as its own through its diversion dam. ⁽¹⁰⁾ Rovella was then functionally a branch of Favara, but it received its customary quota of water and maintained its autonomy as a separate community of irrigators. At an earlier stage of development it might have merged with Favara completely, especially if the canals had been administered by a central authority. Still it is difficult to understand how the lost canal could have been absorbed into a neighboring system without that system's quota of filas being increased -- without, that is to say, doing violence to the concept of proportionality. Most probably, the depopulation of its service area and the accompanying deterioration of the dam and channels would have [213] impelled the remaining irrigators to attach themselves to a more prosperous, efficient delivery network. ⁽¹¹⁾ In any case, the disappearance of the lost canal's diversion dam and absorption of its service area must have antedated the Reconquest; the neighboring canals are documented from the early fourteenth century, and any such major change would have attracted the attention of local authorities.

The true nature of the Valencian fila has puzzled scholars and engineers for generations. Since the seventeenth century attempts have been made to gauge it in hopes of discovering some uniform and concrete measure which would express the volume of water in terms of its flow through a known orifice at a constant velocity, in the same manner as other modern liquid measures.⁽¹²⁾ Such efforts have of course been predestined to failure so long as the proportional concept was not understood.⁽¹³⁾ The adoption of a fixed value for the fila in some areas (such as Murcia and Orihuela) and the continued use of the unspecific, proportional concept in others (Elche and Novelda) caused nineteenth-century scholars like Sir Clements Markham even more confusion. ⁽¹⁴⁾

No less puzzling was the pluma (quill, feather), one hundred forty-four of which comprise a fila. Surely this measure too has the same rationale as the fila: just as a river was divided into so many days of water, expressed in filas equivalent to hours, the fila itself was subdivided according to the same principles. The pluma would be analogous to the fila in an internal turn, one pluma being equal to one hour of water in a channel carrying one fila. This would suggest that internal turns also were conducted on a weekly basis, with no irrigating on the Sabbath.
 

The irrigation measurement unit based on hours is not unique. In widely separated Islamic countries it is standard. In southern Iraq the base unit of water measurement is the waqt (time), which means either the period from sunrise to sunset or that from sunset to sunrise - approximately twelve hours. ⁽¹⁵⁾ [214] In the Yemen (Sho 'ûb) the unit is the fard, which equals twenty-four hours. ⁽¹⁶⁾ The twelve-hour base is also observable in the Ghûta of Damascus, although as in Valencia not all divisions of the river were easily reducible to it (see, for example, the division of the Barada River in the eighth century, Appendix 8). There the standard water unit is now the qîrât, a proportional measure which, though it varies in value from place to place and canal to canal, has the basic connotation of one twenty-fourth. At Dârâya a qîrât has the value of one half-hour of irrigation, although in most other places it means one hour. At Berzé, the value of the qîrât is fixed by local option at one-fourth hour in dry years and one-half in wet. ⁽¹⁷⁾ This procedure is the same as that of "doubling the water" in Spain, whereby the amount of water is doubled and the amount of time halved.

The most striking characteristic of the Ghûta measurement system is that the river is not considered to hold a fixed sum of qîrâts, but at each point along the course where there is a derivation the debit is considered as divisible into twenty-four qîrâts. The various canals then draw from the river a proportional part of the debit at that point (the Yezid Canal diverts eighteen qîrâts -- that is, eighteen twenty-fourths of the river's volume at its dam -- Thûra fourteen, Bânâs twenty-three or almost the entire volume at that point). ⁽¹⁸⁾ Each canal is also considered to have flowing in it twenty-four qîrâts, which are then divided proportionally by means of divisors (called mezzaz) both functionally and in form identical with those of Valencia. ⁽¹⁹⁾

The relation between the concepts of division in the Valencian huerta and the Ghûta of Damascus can now be appreciated. The fila and the qîrât are cognate measures. The Guadalaviar at each of four stages below the Moncada Canal was considered as having twenty-four units. The Syrian model was that which guided the Muslim irrigators of Valencia as well as of several nearby huertas, those of Castellón and Gandia in particular. ⁽²⁰⁾

The Syrian model was not the only one followed in the medieval kingdom. The measurement units of Elche are so similar to those of Sho'ûb that the irrigation system of Elche can truly be characterized as Yemenite. Both the fard and the hilo [215]

Sources: Rossi, "Irrigazione nel Yemen," p. 352; Ibarra y Ruiz, Reigo de Elche, p. 156

To people without mechanical clocks, time is like a distance traveled or a cycle of recurring events. ⁽²⁵⁾ In the irrigation systems studied in this volume the first step toward institutionalization may have been the imposition of a specific cycle, or order, of distribution. Initially -- that is, after demand for water became great enough to occasion the need for control -- the river may have been divided so that all the water went to one canal for a specific amount of time and then to another. The greater the number of canals, however, the greater the hardship likely to [216] fall on any one canal on the days when it does not have water. Furthermore, water could easily be wasted if on the day of its turn a given canal could not use all that was available. This initial stage in which the river itself was subjected to a turn was modified, therefore, so that instead of each canal receiving the total flow for a specific time, continuous flow was given to all the canals simultaneously but in such a way that the proportions of the original division were preserved. For this purpose carefully measured diversion darns and divisors must have been constructed.

The repeated divisions of the Guadalaviar in the fourteenth and fifteenth centuries (expressed in a variety of units including filas, mulnars, and days) seem to indicate an extended phase of transition from the concepts underlying the first stage to those underlying the second. The second, modified system is the more efficient, but it is harder to control than the first. Significantly, the greater the pressure for water in medieval Valencia, the more likely it was that arrangements for distribution would revert toward the archaic method of dividing the river by days. The division of the Mijares in 1346 combined both concepts. The significance of the term "thread" of water may be that it expressed this concept of continuous flow. ⁽²⁶⁾

1. Pul Eliya: A Village in Ceylon (Cambridge, Eng, 1961), pp. 164-165.

2. Sentencia arbitral del serenísimo infante D. Pedro (1346), in Traver, Historia de Villarreal, pp. 472-477.

3. Carlos Sarthou Carreres, Datos para la historia de Játiva, 3 vols (Játiva, 1933-1935), I, 297.

4. J. Musso y Fontes, Historia de los riegos de Lorca (Murcia, 1847). p. 34.

5. Foistavella, Huerta de Gandia, p. 79.

6. "Antiguedades de Elche," p. 336, cf. Markham, Report, pp. 51-52.

7. Sentencia arbitral (1346); see n. 2, above.

8. Ruiz-Funes (Derecho consuetudinario, p. 141) states that the turn does not function on Sundays in Murcia. Semitic law generally did not permit irrigation on the Sabbath,

9. Provincia de Valencia, p. 294, n. 484

10. Aymard, Irrigations du midi de l'Espagne, p. 22, Jaubert de Passa, Canales de riego, II, 217n.

11. Martínez Aloy (Provincia de Valencia, pp. 287-288) suggested that the canals of Quart, Tormos, and Mislata were of more recent construction than the other main canals of the huerta because their names are not found in early privileges (particularly that of James II, Apr 24, 1321). He was unaware that the canals in question are mentioned in the minutes of the city council of Aug 21, 1313 (AMV, Manuals de Consell, I, fol. 91r). Such reasoning is doubtful in any case because major changes in the system of proportionality would have merited royal attention.

12. The history of these attempts is given by Borrull (Tratado, pp. 7-10) and by Manuel María Azofra, Memoria sobre la exacta medición de agua corriente por medio de muelas, filas y plumas (Valencia, 1844).

13. Aymard, the first scholar to understand that proportionality is at the heart of the Valencian distribution system, was adamant on the futility of trying to figure out what the fila was in hard measurements (Irrigations du midi de l'Espagne, pp. 23-25). Chap. vi of Andrés Llaurado's Tratado de aguas y riegos, 2 vols. (Madrid, 1884), I, 83-94 ("Distribución del agua -- sistema distributivo fundado en la proporcionalidad") is based on Aymard. One of the few modern writers to appreciate the concept of proportionality was A. Guillén; see El tribunal de Aguas, esp. pp. 44-45, following Brunhes. See also Exposición de distintas actuaciones, pp 19-22.

14. Markham's Report, pp. 64-65, points out that even when fixed measures are adopted they are not necessarily the same in each place, that of Gandia differing from that of Murcia, for example.

15. Robert A. Fernea, Shaykh and Effendi (Cambridge, Mass, 1969), p. 125

16. Rossi, "Irrigazione nel Yemen," p. 352.

17. Tresse, "L'Irrigation," pp 479--480. In origin the qîrât was the counterpart of the siliquas, 1/24 of the gold solidus. By extension it came to mean the 24th part of anything.

18. Tresse, "L'Irrigation, pp 511--512, and Richard Thoumin, Géographie humaine de la Syrie centrale (Tours, 1936), p 55 rl. 2

19. Tresse ("L'Irrigation," p. 475) derives mezzaz from mazza (to suck). It is more likely derived from máza, to separate (see Appendix 8, n. 4), a usage parallel to almatzem (Gandia), derived from qasama, to divide. For the form of these divisors, compare the pictures in Tresse (pl. 4A) with those of similar divisors in Gandia (Fontavella, Huerta de Gandia, photo 20).

20. The Gandia system is extremely complicated and its rationale is not entirely clear. The river has 48 filas at the Casa Fosca dam, just like the Moncada Canal in Valencia. Unlike Valencia and the Ghûta, however, the lower diversions of Gandia diminish in filas rather than start each stage anew with 24. Finally, the canals of Gandia and Oliva, containing 14 and 21 filas respectively, are assigned arbitrary quotas of hilos in order to make subsequent divisions in a neater fashion By mathematical juggling, however, the final hilo at the lowest diversion has the same value as the original fila (see Fontavella, Huerta de Gandia, fig. 11).

21. Compare Ibarra y Ruiz, Riego de Elche, p. 201; the hilo of the Marchena Canal is always "double" and therefore runs for 6, not 12 hours. In the Marchena system, the cuarta is equivalent to 1 1/2 hours.

22. See Chap. xii, n. 29. Tumen, the 1435 document form, is closer to the Yemenite pronunciation than the usual castilianization azumbre.

23. Markham, Report, p. 57; cf. González Palencia, "Riegos de Veruela," Arabic doc. pp. 79-80, trans. 80--83, the sale of "one-eighth of water" ("althaman wâhid min al-mâ'"). Gonzalez Palencia states that the eighth is equivalent to 3 hours of water (p. 81, n.6)

24. Musso, Riegos de Lorca, pp. 34-36, cf. also the very similar system of Librilla (Murcia) in Ruiz-Funes, Derecho consuetudinario, pp. 178-179.

25. E. R. Leach, "Primitive Time-reckoning," in A History of Technology, ed. Charles Singer, 4 vols (Oxford: Clarendon Press, 1954), I, 112 The cyclical notion is especially apparent in Arabic expressions for turn (e.g. daula, nûba, daur) which connote revolution.

26. The main lines of this hypothesis were worked out in discussion with Professor Arthur Maass.