[349] I. Sources and estimates of wheat consumption
 
A. Sources for the totals in Table 6.1
 
1561: Given as 100 fanegas per day in Manuel Espadas Burgos and Maria Ascensión Burgos, "Abastecemiento de Madrid en el siglo XVI" (1962), p. 110.

1599: From a report of a decision by the Pósito of Madrid to take charge of the bread supply of the city and buy 180,000 fanegas of flour for the year. AVM, Secretaria, sig. 2-96-1.

1608: Based on various documents on the daily and weekly entry registry, and production of wheat and bread in Madrid. These indicate that the government handled more than 800 fanegas per day. AVM, Secretaria, sig. 1-455-2; AHN, Consejos, Sala de Alcaldes, libros, año 1614, fol. 350.

1614: The total is based on the decision to guarantee that the Pósito would be able to distribute 1,000 fanegas of wheat daily during the period of scarcity. AHN, Consejos, Sala de Alcaldes, libros, año 1614, fol. 350.

1628: Based on estimate of 1,000 fanegas as a daily minimum in addition to the supplies of private households, convents, the palace, etc., for a daily total of 1,200 fanegas. AHN, Consejos, leg. 51438-7; and notes made available by Prof. Antonio Domínguez Ortiz.

1630-31: This figure is based on two complementary sources:

a. The registry of bread entering Madrid from surrounding villages for three days in September 1630, with daily totals of 1,121, 1,031, and 1,018 fanegas, exclusive of bread manufactured within the city. AVM, Secretaría, sig. 2-99-4.

b. Detailed accounts of the efforts of the city government to supply the entire city after the bad harvests in the two Castiles in 1629 and 1630. The summary lists the very high total of 1,034,113 fanegas delivered to Madrid. Assuming that this is the total for two years, this [350] suggests a minimum annual consumption of 517,051 fanegas, or 1,420 per day. Allowing for baking within the city, this coincides with source (a). AVM, Secretaría, sigs. 2-102-8, 2-140-6.

1667: This total is based on a list of the bread suppliers and bakers of the city, which suggests a maximum output of 1,025 fanegas per day. The same source indicates that the surrounding towns were obligated to supply 46,454 fanegas of bread per year. AVM, Secretaría, sig. 2-190-5.

1767: These figures are taken from an official report to the effect that the bare minimum necessary was 1,200 fanegas per day, while normal consumption, as in the spring, was 1,500, and the highest levels were around 2,000 fanegas in late summer. AVM, Secretaria, sig. 2-122-1.

1779-80: Based on monthly reports of grain entering the city, with detail on types of purchase. AHN, Consejos, leg. 6775-3.

1784: Based on a report which concludes that the output of the panaderos supplying the city was 2,000 fanegas per day, while maximum consumption reached 2,233 fanegas per day. AVM, Secretaría, sigs. 2-126-7, -9, -22.

1789: AVM, Secretaría, sig. 4-5-67.

1792: AHN, Consejos, leg. 6780-18.

1797: Based on weekly reports of wheat dispensed from the Pósito at a time when it was apparently the sole source of supplies. These averaged about 2,510 fanegas of wheat, equivalent to 2,800 of bread. These registers exist for many years, but only in a few cases is it possible to know what share of total consumption they represent. For 1797, the totals are high enough to preclude other sources of any note. AVM, Secretaria, sigs. 1-131, 1-126, 2-135, esp. document 17.

1812: From a report on the bread supplied to the city, dated July 8, 1812. At that time daily supplies were about 1,367 fanegas, which was considered barely adequate, but contrasted sharply with the 618 per day of the previous June. AVM, Secretaría, sig. 2-137-3.

1815: Derived from a report of the bread-baking facilities of the city and estimated daily capacity of 1,697 fanegas. AVM, Corregimiento, sig. 1-87-21.

1818 and 1820: Manuel Espadas Burgos, "Abasto y hábitos alimenticios en el Madrid de Fernando VII," p. 257, citing AVM, Secretaría, sigs. 2-138-20 and 2-138-42.

1824 and after: Pascual Madoz, Diccionario geográfico, vol. 10, pp. 1015-1072, various tables.
 

B. Comparisons between normal and minimal consumption
 
The estimate of long-term normal wheat consumption illustrated on Figure 6.1 depends on contemporary statements about the difference between normal and minimum or crisis-level requirements. Since the latter type of figure [351] is all we have for earlier periods, it was necessary to establish the difference between the two, and the assumption that a year with minimum or crisis-level consumption represented 70% to 80% of a normal consumption level is based on three calculations:

1. In 1767, it was reported that the minimum was 1,200 fanegas against a normal level of 1,500. Thus the minimum here is 80% of normal.

2. In 1824-29, the average consumption was 682,164 fanegas, while the lowest year, 1824, was 522,343, or 77% of the average.

3. In 1839-47, the average of the available totals was 705,621 fanegas, while the lowest for the bad year of 1847 was 491,453 fanegas, about 70% of the average.
 
II. Serial data on consumption of commodities
 
A. General comment on the use of municipal tax data
 
Many sections of this book depend upon analysis of economic trends as indicated by taxes collected on various commodities, market transactions, and similar types of activity, but it is often difficult to use municipal tax data as they appear in the sources. Municipalities farmed out their taxes with contracts lasting one to six years; this implies an unknown margin of profit for the contractor, and a corresponding difference between figures shown in municipal accounts and the revenue actually collected. Accounts were figured yearly, and thus tax data seldom reflects seasonal changes. When the rental contracts were relatively long, they did not reflect year-to-year economic changes except when the contracts were renewed, or when crises forced adjustment of the terms. These limitations are offset, however, by the fact that municipal taxes reflect long-run economic change and, when contracts were short, reflect shorter cyclical fluctuations in the economy. We must, however, take note of possible complications, including fraud, bankruptcy of tax farmers, evasion, arrangements with wholesalers and transporters, arbitrary exactions by collectors, and the tendency for authorities to settle for negotiated revenue levels. We cannot do much to the figures to compensate for these potential distortions, but it is important to note that the figures themselves are less secure than they may appear.

Another problem, especially with taxes calculated as a percentage of the value of the merchandise, is the degree to which the nominal yield of the tax reflects changes in price rather than changes in the volume of the commodity taxed. To avoid this problem where possible, the tax figures used in Chapter 6 are for duties that were collected at a fixed monetary rate per unit of the commodity. The nominal value of such taxes was thus tied to the actual volume of goods taxed. Elsewhere in the text, including Chapters 2, 10, 11, and 12, it has been necessary to use taxes collected entirely or partly ad [352] valorem. In those instances it was necessary to make tentative adjustments based on available price indices in order to find a better approximation of the real trend of commercial activity. The following sections of this appendix detail the sources and some of the numbers on which the summary calculations presented in the body of Chapter 6 were based.
 
B. Sources on wine consumption
 
1. Long-term wine consumption estimates

In Chapter 6, section I, B estimates are based on a combination of tax revenues and actual consumption totals. The archives provided official figures for the volume of wine consumption for a number of years. These could be used to test the estimates developed from tax yields. The sources for actual volume are:

1621: AHN, Consejos, Sala de Alcaldes, libros, año 1621, fol. 270.

1630-32: AVM, Secretaría, sig. 3-231-1. For 1631, see also AHN, Consejos, Sala de Alcaldes, libros, a fío 1782.

1635-36: MM, Secretaría, sig. 2-232-1.

1638: AHN, Consejos, Sala de Alcaldes, libros, año 1781.

1698-99: AHN, Consejos, leg. 7222.

1731-36: AVM, Contaduría, sig. 4-107-4.

1757: Malilla Tascón, "El primer catastro de la villa de Madrid," pp. 465-466.

1772-1847, various years: Madoz, Diccionario geográfico, vol. 10, pp. 993-1036. For 1772-74, there is a close fit with tax figures in AVM, Contaduría, sig. 3-326-1; and the total for 1829 coincides with AVM, Contaduría, sig. 3-191-4.
 
2. Wine tax revenues and volume estimates

The seventeenth-century wine consumption estimates given in Chapter 6 were derived from the revenues received by the city from a small wine tax imposed at the end of the sixteenth century. Between 1583 and 1680, the wine consumed in Madrid was subjected to 13 different taxes. The oldest, the one used here, was also the simplest and was collected at the rate of 2 maravedises for each of the 8 azumbres in an arroba or cántara of wine. As a result, it was always collected at the rate of 16 maravedises per arroba, unaffected by changes in price or alterations in the size of the azumbre, as was the case with some later sisas. This history of these taxes and their variants is presented in some detail in an eighteenth-century report found in AVM, Secretaría, sig. 2-218-13. Throughout the period in which this sisa was traced, there are clear references to the tax rate and to the tax as a separate item in the contracts and accounts. The rate was not affected by the radical inflations and deflations of the coinage, and the nominal value was tied to the volume of wine, whether [353] collected in devalued or revalued maravedises. Thus, aside from fraud and the profits of the tax farmers, the principal variable in determining the value of the sisa was the volume of wine sold in Madrid. In compiling this series, the annual revenues from the sisa were recorded for as many years as possible for the period before the practice of letting four-year contracts began. Time, and the state of the archives, dictated that only scattered sisa values were collected for subsequent years.

Modest tax reforms were begun during the reign of Charles II, and in the first decades of the eighteenth century the system of sisas was considerably reformed. Madoz, in his Diccionario geográfico, vol. 10, pp. 1006-1007, details these changes in a brief history of the sisas of Madrid. When the series on wine is again fairly complete, it appears in the accounts as one large tax of about 9.8 reales per arroba with two smaller supplements, the first and second quartillo (quarter-real) en arroba. Since the series for the first quartillo is the most complete, that is the one used as the basis for our eighteenth-century volume estimates. Again the conversion is straightforward, since if the tax was one-fourth real per arroba, an estimate of volume could be obtained by multiplying the tax yield by four. While the accuracy of specific figures obtained this way can be challenged, the technique establishes the magnitude of consumption and its trends through almost 75 years.

With the exception of 1681-86 and 1698-99, all tax figures used below come from AVM, Secretaría, sigs. 3-229, 3-230, 3-231, 3-233, 3-234, 3-254, 3-255, 3-256, 3-257, 3-262; and Contaduría, sigs. 1-109-1, 1-110-1, 1-496-2,1-422-1,1-144-2,2-119-1, 1-172-2,3-326-1,and 4-107-4. For 1681-86 and 1698-99, they come from AHN, Consejos, leg. 7222.

The averages given in Table D. 1 are derived from the yearly consumption totals that were developed, based upon manipulation of the sources along with such official figures for actual consumption as the documents provided.

While some of the differences between estimated and recorded volume are considerable, most of them can probably be explained by differences in the twelve-month period actually used. For some purposes, January to December was used, for others July to June. The estimates for 1766-67 are less firm, since they are based on a rough conversion of the larger wine tax to arrobas, using the ratio found for the 1770's. If anything, the figures are high, since that ratio appears greater in later situations.

The test for the reliability of the estimates is simple--dividing the recorded figures by the estimates to find the margin of error between the two. For the early period, the recorded volume of consumption was below the tax-derived estimate as follows:
 

1631	2.0%
1632	15.8
1635	8.5
1636	1.1
1637	2.3
1639	2.4

[354] The figure for 1632 is high, but the accounts make it pretty certain that there is a six-month overlap on the two figures, with the higher total reflecting a twelve-month period including the new harvest of that year, while the lower figure reflected a bad harvest in the preceding year. For most of the eighteenth-century cases in which similar comparisons can be made, the differences are small enough to suggest the same original source for tax and consumption figures.

No great claims are made for the literal accuracy of these figures; the trends are more important. In some cases the figures may include only wine actually taxed, while in others the recorded volume includes exempt wine as well. The years 1630-40 include all types, as do also the much lower figures for 1770-80. Madoz believed that large quantities escaped the tax and were unrecorded in the 1840's. He estimated that the 580,000 arrobas of the time really represented 700,000-850,000. For all of these problems we see no way to refine the data further, and we must proceed on the assumption that in any given quarter-century or so the problems are not great enough to obscure the basic tendencies.
 
C. Sources on olive oil consumption
 
1. Sources for Table D.2

The sources for Table D.2 are generally the same as for wine in the sixteenth and seventeenth centuries. The eighteenth- and nineteenth-century totals and estimates rely on AVM, Contaduría, sigs. 2-119-1 and 3-191-4; Madoz; and Manuel Espadas Burgos, "Abasto y hábitos alimenticios," pp. 274-275.
 

2. Oil tax revenues and volume estimates

All of the figures from 1584 to 1648 are based on the sisa ordinaria del aceite imposed at the same time as the first wine sisa, at the rate of one maravedí per panilla. There were 100 panillas in an arroba, and 34 maravedises in the real, so the conversion from tax figure to volume estimate requires that the tax figure be divided by 2.94. There are no recorded volume figures available for this period to test for the accuracy of this process.

The estimates of the later eighteenth century are based on the single oil sisa of the period. The actual yield of that tax was 5.808 times the estimated number of arrobas. This ratio was found by dividing the total amount of oil for the five years for which we have recorded totals (1780-84) into the total revenues for those years. Two sets of consumption figures for those years were [355] [356] one in Madoz and the other in the municipal archives, with somewhat different yearly totals. The five-year total for the archival data, however, was 99.98% of that for the Madoz data, indicating a common source for both.
 

C. Sources for estimates of meat consumption
 
1. Conversion from number of cattle to pounds of meat

To establish total consumption in homogeneous figures, it was necessary to determine average weight of meat from the different types of cattle. The averages are shown in Tables D.3 and D.4. Accordingly, in the conversions from animals slaughtered to pounds of meat, we assume that a sheep yielded 21 pounds of mutton, a cow 350 pounds of beef, and a calf 73 pounds of veal.
 
2. Explanation of annual totals not given as such in sources

1601: AVM, Contaduría, sig. 3-588-5. Sources detail a full week's consumption for February 17-22 as 63 cows and 1,344 sheep. Using the above conversions, multiplying by 52 yielded the estimate given.

1607-08: The documents associated with the sisa records already cited indicate that between April 1, 1607, and February 14, 1608, 138,752 sheep were slaughtered, an average of 3,016 per week over 46 weeks. Over 52 weeks, this gives 158,519 head. Supposing the same proportion of sheep to cattle as in 1757, this gives 6,171 beef cattle as the base for calculation. This ratio of sheep to cattle is not very different from the single week for 1601 cited above.

1632: AVM, Secretaria, sig. 2-231-3 includes a list of the sheep slaughtered weekly for 27 weeks after August 1. This gives an average of 4,345 per week, an estimated 224,940 sheep over the year. With the same ratio of beef cattle as before, this implies 9,257 cattle.

1743: Palacio Atard, Los españoles, p. 298, cites a list of cattle slaughtered [357] during 12 weeks in 1743. This averages 4,998 sheep and 140 cattle per week, or 259,896 and 7,280 head per year respectively.
 

1751: Madoz gives 315,581 sheep and 10,567 beef cattle for the year.

1763: Palacio Atard, Los españoles, p. 299, indicates 311,186 sheep and 9,503 beef cattle.

1766: Madoz indicates 325,000 sheep, 9,000 beef cattle, and 1,807 calves [358] for the year. These look suspiciously like estimates or stereotypical figures rather than actual ones.
 
 

1789: Palacio Atard, Los españoles, p. 299, gives 320,767 sheep, 9,793 lambs, 16,288 beef cattle, and 3,642 calves. Except for a slight discrepancy on calves, this matches the summary of consumption in AVM, Secretaría, sig. 4-5-67.

[359] 1796-1801: Palacio Atard, Los españoles, p. 300, gives 484,024 sheep and 13,204 beef cattle as the annual average for this period. This is higher than the estimate of 313,382 and 14,213 beef cattle built up from partial figures from 1797-98, but the latter was a bad year for food supply according to AHN, Consejos, leg. 6785.

1807: Palacio Atard, Los españoles, p. 300, gives 278,553 sheep and 14,678 beef cattle.

These estimates obviously have to be taken as order-of-magnitude figures, although they do not conflict with available figures on pounds of meat sold. Palacio Atard gives as his principal sources AHN, Consejos, legs. 6783, 6791-5, and 6796; and AGS, Gracia y justicia, leg. 1001.
 
3. The shift from mutton to beef

As Chapter 6 suggests, the proportion between mutton and beef in urban consumption changed radically at the beginning of the nineteenth century. Table D.5 summarizes the shift.

4. Revenue figures supporting meat consumption figures

Our figures on seventeenth-century meat consumption are supported by a series of revenue figures from the sisa del rastro, 1590-1728, which are published in my article, "Madrid y Castilla, 1560-1850," pp. 116-117.

For the eighteenth century, we have comparable figures for revenue from the sisa del carne mayor, 1741-1808. The records show a sisa del carne mayor and a sisa del carne menor, the latter appearing in 1770, probably because of the different nature of the source. Since we are interested in trends, and since the smaller tax regularly ran 8% to 12% the value of the larger, only the larger tax is listed in Table D.6.

Pork consumption was listed separately in the sources, and is given in Table [361] D.7, which indicates the number of pounds taxed between 1753 and 1806 and during the decade 1838-1848.
 

The sources present an interpretive problem in this case, because it is possible that they include a wider range of pork products after 1789. In that year a distinction appears between whole carcasses and pork taxed by weight, while the earlier sources are silent on this detail. A sharp upward trend of consumption is established well before 1789, in any case, so at the worst the rate of increase is somewhat overstated for the years around 1790.
 

III. Relative prices of basic commodities
 
The discussion of relative prices encountered from time to time in Chapter 6 is based on some simple calculations. For each year that prices are available, the price of a pound of bread was divided into the price of one azumbre of wine, one pound of beef, one pound of mutton, and one panilla of olive oil. Thus, for example, in 1556 an azumbre of wine was equivalent to 5.714 pounds of bread, a pound of beef to 2.517 pounds of bread, a pound of mutton to 4.724 pounds of bread, and panilla of olive oil to 1.379 pounds of bread. The prices are derived from Earl Hamilton, American Treasure and War and Prices, with a few at the end taken from Gonzalo Anes Álvarez, Economía e "Ilustración" (1969), "Las fluctuaciones de los precios del trigo, de la cebada y del aceite in España, 1788-1808."
 
IV. Consumption of nonstaple commodities
 
Our sources yielded serial data on consumption or tax revenues linked with a number of commodities aside from wheat, wine, meat, olive oil, and charcoal. These include quite extensive series on wax, sugar, cacao, soap, and fish, two of which begin in the mid-eighteenth century and the remainder in 1770. Actual quantities are given in many cases. For others, the conversions from tax revenue were relatively direct: wax was taxed at the rate of one quartillo per pound, cacao at one real per pound, and sugar at about 9 reales per arroba (25 pounds). Additional sources provided more scattered evidence on fresh versus preserved and salted fish, chocolate, coffee, and distilled liquor. In tables D.8 and D.9 we attempt to summarize this data in support of some of the discussion of these commodities in Chapter 6, and at the same time provide a rough idea of the levels of consumption in the city at various times. Table D.8 indicates either the level of consumption in 1789 or the average annual consumption in 1786-89 as the base = 100 for the indices of consumption in Table D.9, which in turn are the basis for Figure 6.4. In this way, even without the detailed statistics, the reader can reconstruct a rough estimate of the level of consumption for any commodity through much of the century between 1750 and 1850.
 
 

In addition to the consumption data for the limited range of items listed above, the documents contain detailed lists of urban imports for 1789, 1847, and 1848. These sources are based on fiscally inspired records and no doubt suffer from underreporting, but they do provide some insights into changing patterns of consumption between the eighteenth and nineteenth centuries. Some of the contrasts are discussed in the text, but often with a considerable degree of abstraction. To provide substance for the abstractions, Tables D.lO and D.ll indicate consumption figures for 1789, 1746, and 1848 for all commodities that could be compared directly without confusion due to nomenclature or contemporary classifications. Table D. 10 deals with foodstuffs, Table D. 11 with miscellaneous commodities and manufactures.