APPENDIX A—
THE POPULATION OF SPAIN IN THE EIGHTEENTH CENTURY

Three censuses were taken in the second half of the eighteenth century of all the individuals living in peninsular Spain, the Balearic and Canary Islands, and the African presidios. The first dated 1768 and 1769, known familiarly as the census of Aranda, and the second, dated 1787, the census of Floridablanca, were published together in Censo español executado de órden del rey comunicada por el excelentísimo señor Conde de Floridablanca, primer secretario de estado y del despacho en el año de1787, the first in summary form at the beginning of the book. They give total populations of 9,309,804 and 10,409,879, but they do not refer to precisely the same populations.^[1] The second includes the island of Menorca, which did not belong to Spain in 1768. The population recorded for Menorca in 1787 is 28,177. The census of 1768–69 was taken by dioceses, and it did not include the towns that belonged to nullius jurisdictions, primarily those under the military orders.^[2] The census of 1768–69 recorded 16,427 pueblos and 18,106 parishes, that of 1787 records 18,716 pueblos and 18,972 parishes, but the differences were not all due to the omission of nullius jurisdictions in 1768–69.^[3] The advertencia of 1787 implies that its authors believed the additional units represented new towns.

The date 1787 has always been attributed to the second census, but the date should be either 1786 or 1786–87. The original sheets on which the populations of individual towns were recorded are in the library of the Real Academia de la Historia, Madrid. On copying those from the province of Salamanca for this study, I noted the dates on which 175 of them from different regions were signed. The forms were evidently sent out in early October 1786 to the villages.

[1] These figures come from the "Estado comparativo de las dos operaciones," which corrects for an error in the 1768–69 table.

[2] This difference is not mentioned in the advertencia to the census but is noted in the advertencias to the census of 1797.

[3] The census of 1787 lists 520 ciudades, villas, and pueblos of the military orders.

Fifty-three percent of the sample signed their returns before the end of October, 88 percent before the end of 1786, all by the end of March 1787. On the basis of this information, I shall use the date 1786 for this census.

The third census was published as Censo de la poblacion de España de el año de 1797 executado de órden del Rey en el año de 1801, called the census of Godoy. It gives a total population of 10,541,221, living in 21,210 pueblos and 19,186 parishes.^[4]

The advertencias of the 1797 census say, "Although the present census includes many more items than the previous one, it nevertheless does not give a complete return of the population of the kingdom, because the pueblos have not given their reports with the accuracy that one would wish, for they believe the purpose to be an increase in their taxes." Historians have tended to accept this criticism of early censuses and to assume they underrepresent the true populations. Francisco Bustelo García del Real takes this position and proposes an underregistration of between 5 and 10 percent for the census of 1786.^[5] My impression is that it was more accurate than this, probably as good as most present-day censuses.

Prior to these three published censuses the only nearly complete survey of the population of the country was a count of vecinos or lay heads of household, known familiarly as the census of Campoflorido, collected from the different provinces when Felipe V was struggling to readjust the tax burdens at the end of the War of the Spanish Succession. A manuscript record of this census, which gives the number of vecinos in each town in peninsular Spain except for the Basque provinces, has been preserved in a bound volume in the Biblioteca Nacional, Madrid.^[6] The provincial returns are dated between 1710 and 1717, except for those from Navarre, which go back to 1677–79, and the city of Madrid, dated 1723. Historians have usually applied the date 1717 to this census, but a more accurate one is 1712, the date of the royal order for its confection. I shall use this date, as does Francisco Bustelo.^[7] Adding the vecinos reported by the different units gives a total of 1,024,089.

As a source for the demographic history of Spain, the document has obvious weaknesses, which attend any census of this kind from early modern times. Since it is a count of only the heads of household (including widows), to obtain an estimate of the total population one must multiply the count by a coefficient that represents the mean number of persons in each household. Any such number is of necessity arbitrary, for the average household size would have varied from region to region and occupation to occupation, as Part 2 of this book shows. The royal instruction said that nobles, clergy, and legally recognized poor ("pobres de solemnidad") were to be omitted, and widows to be counted

[4] "Advertencias" (1797) says 2,404 pueblos more than in 1786, which would be 21,120, but the total number of ciudades, villas, lugares, and aldeas is 21,210. I assume that these are what were meant by "pueblos." The census lists five other categories of "poblaciones": granjas, cotos redondos, despoblados, and corregimientos, which total 4,253 units.

[5] Bustelo, "Población española," 93.

[6] Biblioteca Nacional, MS. 2274.

[7] Bustelo, "Algunas reflexiones."

as one-half vecinos.^[8] Not all officials followed this order to the letter. Some provinces counted a widow head of household as one vecino, others as a half, and Salamanca evidently as a quarter. Furthermore only Catalonia, Extremadura, and five provinces of Old Castile reported the number of hidalgos (nobles). Some provinces specifically included legally recognized poor, others specifically excluded them, and most did not say which course they took. Members of the clergy and armed forces were systematically excluded, for they did not live in lay households.^[9] A systematic comparison of this census with the later ones indicates that here indeed local authorities reported lower figures than the real ones, probably out of suspicion of the royal intentions. Of course, not all authorities would have cheated at the same rate. In Salamanca province, for instance, underreporting appears to have been much greater in towns under seigneurial jurisdiction than in those under royal jurisdiction. Like the vecino-population coefficient, no correction for underreporting can be more than an educated guess or represent all regions equally well. Finally the manuscript census has errors of addition, although it would take a detailed addition of the individual town returns to determine how many.

Historians have made little use of this document.^[10] Instead they have relied on a publication of the returns in this census by the eighteenth-century economist Gerónimo de Uztáriz in his work Theorica y practica de comercio y de marina (1724). Uztáriz published the vecino count province by province, with a total of 1,140,103 for the entire country. This total does not match the manuscript total for various reasons. The most obvious is that Uztáriz added his estimates for the Basque provinces (he gave them the same figure as that of Navarre in 1677–79!) and Mallorca, Ibiza, and the presidios of Africa. In addition, without informing his readers, he increased certain provincial returns before he published his table. One might suppose that he had a census at hand that was different from the known manuscripts, were it not that his figures represent additions in round numbers, usually thousands, to the manuscript figures, even when the manuscript figures are erroneous totals of the individual town returns. Thus the manuscript gives totals of 14,974 vecinos for Guadalajara province and 25,556 for León. Uztáriz's figures are 16,974 and 28,556 respectively, and the manuscript total for León is incorrect in the first place.^[11] The reasoning behind his corrections defies analysis. There is no obvious geographic pattern. In one-third of the provinces, all in the northern half of the country, he preserved the original census count. In the rest, his additions ranged from 5 to 13 percent, except for Asturias (49 percent), Jaén (36 percent), Madrid city (23 percent), and Valencia (whose total he reduced inexplicably by 6 percent).

Although Uztáriz did not inform his readers of these changes, he described attempts he had made to check the accuracy of the returns in Madrid province and in Andalusia. They led him to conclude that a fair approximation was that

[8] Ibid., 98 n. 20.

[9] See Uztáriz, Theorica, 37–38.

[10] One who has studied it is Bustelo, "Algunas reflexiones." He has seen a second manuscript copy of it in the Biblioteca Menéndez Pelayo, Santander.

[11] The total of the figures for the fifteen partidos of the province is 25,513.

the census overall represented only 80 percent of the real count. His corrected figure is 1,425,000 households. He made further additions to take into account the military forces and their dependents (whom he calculated to be 180,000 persons), clergy and their households (250,000), foreigners resident in Spain (40,000), and 30,000 shepherds who he believed never got reported. His corrections conveniently total 500,000 persons. Uztáriz estimated 5 persons per vecino, and on this basis reached a grand total of 7,625,000 people in Spain, which he rounded off to 7.5 million to cover any "involuntary excess" in his calculations.^[12]

Most historians have concluded that Uztáriz's figure is too high, either because underreporting would not have been as great as he assumed or, what is more likely, given current knowledge of early modern households, because his coefficient for converting vecinos to population was too large. José Canga Argüelles, an early nineteenth-century official of the ministry of hacienda, accepted his total number of vecinos before any augmentation and multiplied it by 5 to get a figure of 5,700,515.^[13] Canga Argüelles's figure gives a growth of 83 percent between 1717 and 1786, Uztáriz's, 39 percent. Recently Gonzalo Anes, after careful consideration, has considered a 50-percent increase in the population reasonable.^[14]

To avoid reliance on Uztáriz, Massimo Livi Bacci, an expert historian of Mediterranean populations, uses a different approach. He calculates a probable life expectation in Spain in the eighteenth century of twenty-seven years by projecting backward the known trend in life expectation of Spaniards for the late nineteenth and early twentieth centuries. On the basis of this figure and the age structure shown in the census of 1768–69 (very similar to those of the censuses of 1786 and 1797), Livi Bacci calculates from model life tables that the rate of growth would have been about 0.40 percent per year.^[15] Using this rate and projecting backward from 1768–69, one obtains a population in 1712 of 7.43 million, close to the estimate of Uztáriz. This figure gives a 40-percent increase in population between 1712 and 1786 for a mean rate of growth between these dates of 0.46 percent per year. This is very close to the rate of growth for England and Wales between 1711 and 1786, 0.44 percent, based on the estimates of Wrigley and Schofield for these two dates,^[16] much higher than the best estimate for that of France, 0.27 percent,^[17] and above that attributed to Italy, 0.38 percent.^[18]

To compare the rate of growth in the periphery and the interior, one must

[12] Uztáriz, Theorica, 34–39.

[13] Canga Argüelles, Diccionario de hacienda, s.v. "Población que se regulaba a España en los años de 1700 a 1723."

[14] Anes, Crisis, 132–40.

[15] Livi Bacci, "Fertility and Nuptiality," 84–89. Livi Bacci reviews various twentieth-century estimates of the Spanish population in the eighteenth century, which I pass over here.

[16] Wrigley and Schofield, Population History, 528–29, Table A3.1.

[17] The total population figures for France given in Morineau, Faux-Semblants, table on pp. 294–95, produce rates of 0.27 and 0.28 percent, for the period 1700 to 1780 and 1787.

[18] Livi Bacci, "Fertility and Nuptiality," 88.

disaggregate the national figures. Livi Bacci does so for the totals of Uztáriz and the census of 1786, but no clear pattern emerges.^[19] Since, as we have seen, there is no apparent justification for Uztáriz's haphazard changes in the manuscript totals, we may more properly take the original figures for calculating the population distribution about 1712. Furthermore, instead of using the total population in 1787 for purposes of comparison, one can from the census data estimate the number of vecinos in that year.^[20] Table A.1 compares the proportions of total vecinos in each district at the two dates (less the Basque provinces, for which the vecindario gives no figures). It shows that in the periphery Galicia, Asturias, Valencia, and Murcia grew faster than the national rate, while Catalonia grew more slowly. In the interior, Navarre, Aragon, and Extremadura were falling behind (contemporaries believed that Extremadura was actually losing population),^[21] while most of the meseta was growing about as fast as the country as a whole. Andalusia was growing more slowly, except for Seville province. Because the vecindario of 1712 is unreliable, these comparisons cannot be precise, but the trends are probably meaningful. Only a detailed study of all regions can reveal the full extent of the errors.

Catalonia and the villa of Madrid appear anomalous, because the table shows them to be falling behind the country as a whole, whereas the studies of Pierre Vilar and David Ringrose indicate that they were growing rapidly. The fault appears to lie in the data in Table A.1. After a detailed study of the eighteenth-century censuses of Catalonia, Vilar has established that the 1712 vecindario did not underreport its households, as it evidently did elsewhere, and may have augmented them slightly as a punitive tax measure for its opposition to Felipe V in the War of the Spanish Succession.^[22] If this is the case, the 1712 vecindario would give Catalonia a higher proportion of the total population of the peninsula than it actually had. The case of Madrid is similar. The vecindario of 1723 (the date of its survey) was fairly accurate, although excluding clergy, garrisons, and transients, as does the census figure shown for 1786 in the table.^[23] Because of the underregistration of 1712, the figure of vecinos for all Spain in 1786 is 139 percent above that of 1712. Places such as Catalonia and Madrid, where the early figure is nearer the truth, thus show a relative decline where there was none. Their cases indicate the dangers of any attempt to determine relative growth from the sources available. Nevertheless, with all due caution, one may conclude that the periphery of the north and east was growing slightly faster than the interior.

[19] Ibid., table on p. 213.

[20] I use the total of all married men, all widowers and widows under fifty, one-half the widowers and widows fifty and over, and one-half the single men twenty-five and over. Bustelo performs a similar operation for 1797, using all married men twenty-five to fifty-nine, one-half the single men twenty-five to fifty-nine, and one-half the widows under sixty ("Algunas reflexiones," 100–101).

[21] Anes, Antiguo Régimen, 181, citing Antonio Ponz.

[22] Vilar, Catalogne 2 : 18–20. Anes, Antiguo Régimen, 29–35, provides additional information and confirms the rapid growth of Catalonia.

[23] Ringrose, Madrid and the Spanish Economy, 19–33, discusses the population of the city from the sixteenth to the nineteenth century.