A Brief History of Mexican Statistical Practices

When the Mexican Agricultural Society proposed to contribute “a statistical table of the Republic, original and unedited, that presents the state of our agriculture with the greatest exactitude possible given the difficulties that this Secretariat well knows” for the 1900 Paris Exposition, the modernizing farmers and policy-makers behind the project situated it in a long trajectory of contested knowledge production and accumulation in the country.Footnote ¹¹ Founded in 1879, the Agricultural Society was a private promotional organization that brought together bureaucrats (the first honorary president was Mexico's then secretary of finance), agronomists, and planters.Footnote ¹² The society collected scientific findings, distributed market information, and sent endless questionnaires about the state of and prospects for Mexico's countryside, publishing a monthly bulletin to distribute their collective findings. When it came to representing their undertakings on the world stage at the Paris Exposition, specimens and examples of Mexico's raw and industrial products were well and good, but not sufficient. Published data was another key marker of the country's advancement and the way to attract foreign capital and immigrants alike. Along with civil society organizations like the Mexican Academy of Jurisprudence and Legislation, the Geological Institute, the National Museum, and the Mexican Pharmaceutical Society, the Agricultural Society promised to deliver a useful collection of previously published materials as well as a compendium of newly collected data that would describe and promote their country's current prosperity and future promise.Footnote ¹³

The officials in charge of Mexico's pavilion at the Paris Exposition quickly agreed to the society's project, with the caveat that the effort should not replicate work already being done by the country's General Office of Statistics (Dirección General de Estadística), to avoid extra work on the part of those providing the data. Housed within the Department of Fomento, a government ministry that directed projects related to modernization and promotion, the bureaucrats designing and assembling Mexico's contribution to the Paris Exposition asserted that when they received agricultural data from the proposed survey they would then integrate it with data already held by the General Office.Footnote ¹⁴ Even as Fomento officials and Agricultural Society members promoted their new undertaking, they also acknowledged prior data collection efforts, as well as their deficiencies and difficulties. The new survey would overcome past challenges and represent a cohesive and coherent picture of Mexico's agricultural present and future. At the same time, all involved recognized the country's statistical undertaking as a work in progress—a conversation between Mexico City and the provinces in which the vocabulary was not yet settled.

The project of state comprehension and standardization that undergirded the agricultural survey had deep roots. Statistics as an academic and administrative field of expertise was still consolidating in the late nineteenth century, but attempts to catalog and advertise the diversity of New Spain's terrain, climate, populace, and products were commonplace in the colonial era as well. Such representations, whether in paintings, specimen cabinets, or municipal censuses, were used to argue for colonial subjects’ capacity for self-governance, and eventually for their right to it, even as they erased the indigenous knowledge that had facilitated their collection and centered Mexico City and its environs.Footnote ¹⁵

With independence, politicians and bureaucrats joined the global move from natural history to state-led information management in the form of statistics. Whether in colonial holdings or new republics, governments took up statistics as the science of the state (its etymological definition) and sought to deploy the ongoing counting of people and resources in service of state functions. As Leticia Mayer Celis has written, Mexico's nineteenth-century statistical projects, including those mandated by the constitution of 1824, were part and parcel of a broad North Atlantic interest in creating a scientific view of society that could prescribe solutions rather than just describe the world as it was.Footnote ¹⁶ Prussia founded the first central statistical offices in the early 1800s, but Latin American nations were not far behind, with Mexico's Instituto Nacional de Geografía y Estadística established in 1833.Footnote ¹⁷

These early offices existed as semi-autonomous organizations, situated at the intersection of the administrative state and a coterie of private promotional and scientific societies dedicated to creating, accumulating, and disseminating knowledge.Footnote ¹⁸ Latin American governments moved the functions of these disparate organizations, which had long provided them with data, fully inside their state apparatus around mid-century.Footnote ¹⁹ Mexico's statistical office was placed within the new Department of Fomento, or Development, at its creation in 1853. Fomento was a department dedicated to projects that would promote Mexico's economy and its profile, with offices on activities as varied as public works, colonization, scientific exploration, surveying, and publications. Statistical activities were right at home. Fomento's data collection and reproduction activities proceeded in parallel with similar undertakings in the Secretaría de Hacienda (Department of Finance), until the Mexican Congress passed a law founding an independent General Office of Statistics in 1882.Footnote ²⁰ Even as they gained stature, national offices continued to rely on private organizations and local governments for the bulk of their data while attempting to regulate and consolidate the terms and methods of its collection.

From ambitious projects like a national property registry to detailed surveys on the production of wine in each municipality, data collection became a regular means of interaction between provincial producers and representatives of the national government and its projects, whether within the administrative state or as adjunct to it.Footnote ²¹ From mid-century, municipal council minutes regularly recorded the receipt of new surveys or questionnaires along with instructions to the appropriate functionary to find, record, and return the requested information.Footnote ²² Mexico's national archive contains endless boxes of manuscript responses to such surveys, and administrative and organizational publications provide endless typeset transcripts of those pages, and more.Footnote ²³ Even though statistical offices within various branches of government might have aspired to employing their own enumerators, municipal officials and private individuals with a penchant for promotion or study remained the primary source of data into the twentieth century. Transcriptions of the information they sent was republished in the monthly bulletins of organizations like the Agricultural Society or the Department of Fomento and circulated around the country and abroad to society members and government officials alongside articles on innovations in machinery, plant breeding, and markets.

People far from Mexico City willingly participated in the project of state information collection, but they did not always understand or represent their land or themselves in the terms provided by Mexico City technocrats and promoters. Local officials and farmers reinterpreted column headers and survey vocabulary, gently ridiculed Mexico City's expectations of what they were currently producing or could potentially produce, and provided answers that spilled across cells and blank spaces when a single number or word could not capture their reality. The editors of Fomento's monthly bulletins came to accommodate these dissonances, publishing the long-form responses to their surveys alongside or in lieu of tabular data by the 1890s.Footnote ²⁴ Abstraction into the kind of clean dataset harnessed in modern statistics was not always possible or even desirable. Both those requesting data and those providing it recognized that nuance and detail had its value.

At the same time, tabular data and the calculations derived from it proliferated in government publications and the popular press. The concept of gross domestic product (GDP) was still decades in the future, but monetized data on production, yield, pricing, costs and standards of living, and more blunt measures of national wealth were in high demand.Footnote ²⁵ Mexican bureaucrats attempted to calculate riqueza nacional (national wealth) as early as the 1850s.Footnote ²⁶ In 1862 and 1877, respectively, José María Pérez Hernández and Emiliano Busto published extensive volumes of compiled statistics that sought to represent the entirety of the nation's population, production, natural resources, political and social structure, and more.Footnote ²⁷ Hacienda began to publish monthly reports of exports and imports by port and by type of good in the early 1870s, accompanied by summary reports at the end of each fiscal year.Footnote ²⁸ Newspapers published projected state and national budgets and reported on government spending. State governments published their own statistical reports and cadastral surveys, quantifying tax incomes and property size as part of larger efforts to privatize and standardize property-holding.Footnote ²⁹ The year 1895 marked the first time the Statistical Office undertook a national census, a task it had been charged with in 1883. At the time of the agricultural survey, statisticians were still processing their general summary.Footnote ³⁰

Mexico's 1867 public education law is used by many historians to understand the Mexican elite's ideological orientation and favored policies in the subsequent decades. While statistics it was not one of the key disciplines enshrined in the law for the coming decades, it nonetheless met the needs of a nation intent on order and progress.Footnote ³¹ Citing the work of European political economists and statistical thinkers like Schlozer, Sinclair, and Moreau, as well as pointing back to ancient societies from Egyptian pharaohs to the Inca, Mexico's bureaucrats asserted that statistics provided governments with “the only data that can provide solutions to the gravest administrative problems and provide for the prosperity of the people.”Footnote ³² Statistics could connect the most mundane needs to the grandest aspirations of a polity.Footnote ³³

Thus, by the time the Agricultural Society proposed its survey for the 1900 Paris Exposition, data collection, both qualitative and quantitative, was something people across Mexico were accustomed to, even if their adherence to the statistical norms and terminology used in national projects remained unconsolidated. Despite repeated attempts to standardize the management and exploitation of rural property, the Mexican government had minimal control over the vocabulary and the legal reality that governed rural spaces.Footnote ³⁴ The countryside remained incommensurable. Nonetheless, years of data-gathering led higher-ups at Fomento to assert their authority over how the countryside should be represented and to predict the easy completion of the grand statistical project whose results would be carried to the Paris Exposition.

In editing the survey proposed by the farmers, agronomists, and businessmen of the Agricultural Society, officials less familiar with rural realities deleted columns related to products that grew wild, scarcity of labor, transportation routes and prices, and consumption markets, asserting that they could derive this information from what they already had at hand.Footnote ³⁵ At some point between the approval of the draft table and the printing of 6,000 copies to be sent to state governors, columns related to crops in production and their value were cut from the sheet—to my great present annoyance, and very likely the annoyance of whoever initially attempted to compile the data that was sent back. Despite the regular practice of requesting information on all crops in production and insisting that local officials not skimp on details—and despite the recognized value of eliciting detailed responses to questions about cultivation and markets—the officials who made the final version of the survey crafted a table with minimal space for nuance, explanation, or variety. The material structure of the table itself reflected the outward-facing representational project of one set of Mexican bureaucrats, committed to looking and acting as a state, even as it undermined the aspiration of a different set of data-driven officials to see the totality of Mexican agricultural production.

The Unpublished Statistics

The brief description of the agricultural statistics included in the official report on the Paris Exposition put an end to the self-assurance of Agricultural Society proponents and Fomento technocrats. Tucked away in the detailed accounts of their work to collect, transport, and display specimens of the country's handicrafts, agricultural goods, industries, minerals, and artistic and cultural achievements, Fomento officials employed the passive voice to attribute the failure of the agricultural survey to the late arrival of the requested data.Footnote ³⁶ The information the compendium was to intended to contain, including a number of the columns edited out of the circulated tables, was enumerated, but the incomplete project, confounded by poor planning and the predictable unruliness of responses, did not merit more than a few passing sentences.

Although they were the target of much of the blame for the project's failure to achieve its objectives, municipal officials throughout the country did in fact take active part in the project. Fomento circulated some 6,000 printed copies of the blank tables, each measuring about 24 inches by 18 inches, to Mexico's 2,792 municipalities in the spring of 1899. If there was an instruction sheet that accompanied the tables into the hands of the nation's governors, I have not found it.Footnote ³⁷ The national archives contain surveys returned from 1,373 municipalities, together representing 18 of Mexico's 30 states and territories. The surveys arrived in Mexico City across the summer, fall, and winter of 1899, with the last arriving from Guanajuato and Durango in the spring of 1900, far too late to be processed before the Paris Exposition began.Footnote ³⁸ The surveys for Chihuahua, Campeche, Colima, Puebla, Sinaloa, San Luis Potosí, Tabasco, Tlaxcala, Veracruz, and Zacatecas were either never returned or have slipped through archival cracks in the ensuing century.Footnote ³⁹ Similarly, I have never found evidence of anyone returning the tables related to cattle and local markets that were supposedly circulated alongside the agricultural surveys.Footnote ⁴⁰

While no state has a complete set of responses, there are reports from at least two-thirds of the municipalities in every state that did submit forms.Footnote ⁴¹ It is difficult to discern any patterns in the municipalities and states that did not return forms: a number of mining towns, urban centers, and even Mexico City's military quarters (cuarteles) did respond, sending in tables indicating they had minimal agricultural properties, but many highly agricultural regions returned no information. Some of the states whose forms are missing were governed by long-term allies of President Díaz, making it hard to assert that any sort of political message was being sent by refraining from returning the tables.Footnote ⁴² I would have a difficult time attributing political meaning to the blank spaces on the map of my dataset, given the known archival fault lines that run through it.

Instead, it is within the returned tables that Jessica Marie Johnson's idea of the null value comes into the play. The variety of ways in which local officials made decisions about providing information (or not providing it)—which column headers to ignore, which to recalibrate, which to accept—provides insight into the continued incommensurability of definitions of agricultural production in turn-of-the-century Mexico.

Municipal presidents or secretaries (or both) took charge of and signed most tables, generally abbreviating their titles and affixing their municipal seal to the finished forms. In Oaxaca, Fidel Sandoval, chief of the Second Section of the Interior Department, signed all the tables himself, after remaking their headers and column labels to better summarize the state's 1,117 municipalities in as little space as possible. A few districts in Chiapas, Coahuila, and Guanajuato returned their forms unsigned, while most officials in Hidalgo, Michoacán, Morelos, the Yucatán, and part of Guanajuato did not include their titles. Differences in handwriting between the information in the tables and the signatures suggest that the signatory did not always fill out the form, but he (always ‘he’) likely oversaw the process by which data was accumulated. A very few municipal officials indicated that they drew on local tax registers or cadastral surveys as they filled in the form, while a few others wrote that they surveyed property owners directly.Footnote ⁴³ Others admitted their responses were estimates or best guesses.Footnote ⁴⁴ With no place for respondents to include information about sourcing, I can only speculate as to whether such documents, even with the trove of prior statistical exercises conducted at the request of national and state governments, provided most of the requested data.

While the data does not provide complete coverage of the Mexican countryside, it is a more extensive, more detailed, and (at least in theory) more commensurate accounting of Mexican agriculture in this period than we have yet had access to. This was the promise made by the Agricultural Society when it proposed the project: by asking everyone the same questions at the same time, the “original and unedited statistics” would overcome the diversity and proliferation of detail that had both encumbered and enriched prior attempts at a statistical portrait of Mexican agriculture. We as scholars, though, must be cautious in our enthusiasm for such a project, in much the way that Fomento officials themselves realized.

In transcribing thousands of pages of surveys, my research assistants and I have come to see that the incommensurability and seeming incompleteness of the project reveal as much as any statistical analysis could offer. The Agricultural Society and the Department of Fomento may have laid the blame on recalcitrant or tardy municipal officials, but comparable statistics on the Mexican countryside remained a relative impossibility so long as the people surveyed continued to understand their livelihoods in diverse and often dissonant ways. By filling out the tables to reflect their own experience of rural productivity, municipal officials and local producers confounded the state's project of consolidation and authority. But this apparent dissonance does not reflect rebellion or an outright refusal to cooperate; the data returned by local officials, rather than obfuscating or dissembling, demonstrates an ongoing insistence on understanding and depicting local agricultural and economic activity beyond the terms set by the state.

Defining, Delineating, and Defying Property

Property was the organizing principle behind the layout of the agricultural survey. The first column designated the unit for which all other information would be entered. It was labeled “Nombre de cada hacienda o rancho” (Name of every hacienda or rancho). This, of course, asserted and assumed a particular kind of property as the unit of production for the Mexican countryside. In an era when desamortization, surveying, titling, and other modes of promoting private property predominated in public discourse around land use, it is not surprising that a document produced by the central government presumed titled, named properties. It should also be unsurprising, given a growing body of scholarship to this effect, that not everyone in rural Mexico held private title to their land or believed that only privately held land might be seen as productive.Footnote ⁴⁵ The concept of consolidated property regimes was enshrined in the survey's column headings and national cadastral projects alike, but rural producers continued to remind Mexico City technocrats that their project of legibility and standardization was incomplete. Even as they readily provided measures, names, and values, municipal officials undercut Mexico City definitions of productivity and property.

Privatization and registration legislation across the nineteenth century led to a varied rural landscape, as municipal governments, foreign surveying companies, wealthy landowners, and villagers all made their mark on the timeline and manner in which law became reality. There were places where large landholders controlled the process and benefited hugely from the possibilities the law presented for scooping up nearby village lands. Public lands surveyed and sold by concessionaires tended to end up in the hands of hacendados. At the same time, land companies had little interest in grappling with the complexities of village holdings and gave them a wide berth.Footnote ⁴⁶ Mandates to divide and sell villages’ communal holdings also rarely managed to overturn local understandings of who held what.Footnote ⁴⁷ Privatization laws included clauses providing lands for a minimal fee, or no fee, to those of little means, so even where titling did occur, it did not necessarily mean the displacement of those already working the land.Footnote ⁴⁸ All in all, despite liberals’ celebration of private property and its potential for generating economic transformation in Mexico, the landscape remained a motley quilt of property regimes into the twentieth century.

Setting aside the variation in property regimes, municipal officials’ reporting on their local landscapes further countered any projection of uniformity from Mexico City. The 1899 agricultural survey provides one of the clearest examples of this persistent refusal to accommodate national standardization. The table in Figure 3 shows the number of properties reported by municipality, but no municipal official included all properties in his table, and officials differed in how they determined what they should report on.Footnote ⁴⁹ The “haciendas y ranchos” part of the table header lacked official definition, and while states in the center of the country—Aguascalientes, Michoacán, Morelos, Hidalgo, Tepic (now Nayarit)—tended to add the property type in parentheses or include it as part of the property name, such labeling was uncommon elsewhere.Footnote ⁵⁰ Most officials did take seriously the request for property name, with almost all rows containing a saint's name (there were 158 San Antonios and 99 San Josés), an aspirational name (37 La Providencias, 44 El Refugios, and 24 La Esperanzas), or a descriptor (26 Buena Vistas, 22 Rancho Nuevos).

Figure 3 Number of Properties Reported by Municipality. Prepared by author, 2023. Interactive version of this map available at https://caseylurtz.com/agricultural-statistics.

An analysis of the full dataset demonstrates that each official had a relatively consistent practice of selecting properties for inclusion on the basis of value and size. About half included only properties over a certain size and value—with an extension of at least 100 hectares, and a value of at least 500 pesos being a fair generalization—while the other half threw such definitions to the wind and included a wide variety of properties they saw as agriculturally productive (see Figure 4). In its totality, the dataset makes clear that officials in the center and south of Mexico were more inclined to include smaller properties of lesser value, while northern officials tended to note only large ranches and latifundia. Everyone, though, included a wider variety of property types than the historiography might lead us to expect.

Figure 4 Average Property Size and Value, Summarized at by Municipality. Prepared by author, 2023. Interactive version of this map available at https://caseylurtz.com/agricultural-statistics.

The 250 or so municipalities that returned blank or sparsely annotated tables provide additional insight into how officials made their decisions about what properties to include. One respondent, Franco Urbino, the municipal president of Chanal, Chiapas, explained his blank form by writing that “there are neither ranchos nor haciendas as all the indígenas work in the town's ejido.”Footnote ⁵¹ Urbino judged that the ejido, one of the communally held and managed lands the national government had been trying to privatize for half a century, was not the kind of productive landscape Fomento cared about. A few officials reported that mining was their only economic activity, while elsewhere authorities wrote that the land was divided into such small parcels that none could even be labeled ranchos, “por estar tan fraccionado.”Footnote ⁵² The vast majority of such annotations, though, came from Oaxaca. There, Fidel Sandoval from the Interior Department faced down the state's more than 1,000 municipalities by remaking the government-issued document to fit the rural reality as he comprehended it. He created a single table for each district, inserted a new column for recording the name of the municipality, and made liberal use of the phrase “No hay Haciendas ni Ranchos en estos pueblos según se desprende de las noticias respectivas” (There are no haciendas or ranchos in these towns, according to the respective information). Or, more simply, he sometimes wrote “No hay” (there are none). The 243 municipalities Sandoval listed at the bottom of each district sheet constitute the bulk of all places that reported no information.Footnote ⁵³

Sandoval's reasons for failing to report haciendas and ranchos were contradicted in other parts of Mexico. In 81 municipalities distributed more or less equally across the country, officials (unlike Francisco Urbino in Chiapas) included all or parts of the village ejido alongside the named properties. Some 46 officials listed the ejidos as a single unit, as in “the ejidos of [town name],” or simply wrote the name of the town itself, while the rest included multiple individual entries labeled as “fracción” or reported with a unique name. Elsewhere, a process of privatization was implied by the phrase “pequeñas propiedades” (small properties) to cover a number of unnamed lots, which were grouped together in a single row of the table.Footnote ⁵⁴ Even Sandoval, he of the 243 property-less municipalities, included lots as small as 0.35 hectares in his reports.Footnote ⁵⁵ All such parcels were understood to be key parts of the agricultural life of villages throughout Mexico, despite the long history of campaigns to eliminate communal ownership.

While officials from north to south included ejidos in their reports, those in the center and south of the country were more likely to provide information on properties under 100 hectares. That is the figure I use in my designation of a small property, based on norms in the dataset. This division corresponds to established demographic and cultivation patterns: the arid north had a later history of settlement and a greater focus on large-scale ranching and commercial grain and fiber cultivation, while the center and south of the country had settled agricultural towns with longer histories. At least 40 percent of the properties reported in Chiapas, Guanajuato, Guerrero, Hidalgo, Mexico City, and Oaxaca contained less than 100 hectares. That said, Aguascalientes reported details on no properties under 100 hectares, and in Tepic (Nayarit) and Morelos, only 8 percent and 15 percent, respectively, of properties listed were so small. All the northern states reported less than 40 percent of their properties as smaller than 100 hectares, though Tamaulipas included 133 such lots (31 percent of its total). The inclusion of so many small properties, particularly in the center and south, but also some in the north, reminds us of the persistence of subsistence and smallholding even as larger haciendas began to encroach on village lands.

A few brief notes on property size and value are merited here. My decision to demarcate the differences between small and large properties at 100 hectares and 500 pesos is based on both the norms revealed by the dataset I have used and the legal practices of the era, which required public registration of transactions valued over 500 pesos.Footnote ⁵⁶ The overlap between the two markers is not uniform, but in general, municipal officials either did or did not include properties under that size and value. While the average property size for the 14,900 properties in the dataset was 1,642 hectares, the mean was only 143 hectares. Many other scholars use 1,000 hectares or something close to it as the cutoff point for large properties, while Jean Meyer reports the size of an average agricultural holding in Mexico in this era at 5,600 hectares.Footnote ⁵⁷ These figures obscure the variety of agricultural holdings, and the prevalence of that variety, that made up the landscape of rural productivity in turn-of-the-century Mexico, especially as that landscape was understood by those in the countryside rather than those in Mexico City. Other scholars’ high average property size figure feeds on and feeds into the narrative success of privatization projects as projected by Porfirian technocrats and also by Revolutionary-era reformers. who represented themselves as toppling the greedy land-grabbing oligarchy. In 1899, only about half the municipal officials who took part in this project of visualizing the countryside's productivity considered that agriculture happened only on properties over 100 hectares in size and greater than 500 pesos in value.

Municipal officials also had no problem in representing how much of the land remained uncultivated. The table circulated by Fomento nowhere asked for total property size. Instead, it asked for five different measures in hectares: extension exploited, extension not exploited, extension of irrigated lands, extension of rain-fed lands, and hilly or woodland surface area (superficie montuosa).Footnote ⁵⁸ My comparisons of property size are based on the sum of exploited and unexploited land, but it is worth noting that this sum was not always equal to those otherwise calculated.Footnote ⁵⁹ In total, the surveys account for more than 40 million hectares of land spread across the countryside (out of a total national surface area of almost 200 million hectares, or 1.9 million square kilometers). Some 24.5 million of those hectares, or about 60 percent of the total in the surveys were reported as unexploited. And while Mexico's Secretary of Fomento insisted that only irrigated land could be counted as agriculture, only a little over one million hectares (less than 1 percent of the country's total territory) were reported as irrigated, with another 17.25 million listed as rain-fed.Footnote ⁶⁰ Another 21.5 million hectares (about 11 percent of the total territory) were reported as wooded.Footnote ⁶¹

Clearly, the technocrats at Fomento and the Agricultural Society recognized that much land remained unplanted, given their inclusion of these columns in the table. Cartographic representations of Mexico's agriculture from 1885 similarly included “uncultivated land” as a category.Footnote ⁶² To count these lands as unproductive, though, misses the point. Germán Vergara's 2021 summary of nineteenth-century energy regimes explicates how such lands contributed fuel for heating, cooking, and steam power. They provided building materials, afforded foraging that yielded goods for sale and personal use, and supported broader ecologies.Footnote ⁶³ Forests were also seen as necessary for the maintenance of rainfall, a theory environmental historians refer to as dessicationism, and early conservation movements in Mexico resulted from a consensus across social strata that forestland must be preserved. This, perhaps, is why land that could be described as “montuoso” was counted apart, and why its associated calculations do not always square with those of the other columns: woodlands were still recognized as largely a public good, even as privatization laws chipped away at their legal status as communal holdings.Footnote ⁶⁴

By reporting numerous small properties and also regularly documenting large expanses of forested and uncultivated lands, many municipal officials asserted a diversity of productive landscapes against the homogeneity of the haciendas and ranchos implicit in the wording of the column header. They rewrote the table to capture their realities, countering the technocrats’ admiration for plantation systems, as they carved out space to work against homogenizing attempts at legibility. However, about half the officials who submitted tables did adhere to the official rhetoric of rural productivity based on large-scale agriculture. What Mexican officials wanted to advertise abroad had also taken hold at home: for some, agriculture meant irrigation, extensive cultivation, and the consolidation of sizeable holdings. The remaining municipal officials pushed back and included everything from communally held ejidos and half-hectare private plots to the expansive ranches and plantations Fomento officials imagined. To them, it did not make sense to exclude small-scale agriculture on either privatized or communal lands from an accounting of what made the countryside productive. Smallholders paid taxes and produced foodstuffs, usually for regional consumption, which made their land and their use of it meaningful to many local officials. The promotional aims of Fomento officials notwithstanding, the Mexican countryside remained a patchwork of modes of production.

Seeds Employed

In the table, the fourth column from the left is titled in a small font displayed vertically: “Rendimiento por extensión ó semilla empleada” (yield by extension or seed employed). Like the rest of the headings, it implies a shared understanding of agricultural productivity. Data in this column, though, is the least comprehensible of the entire dataset. No matter how much my research assistants, students, and I tinker with numbers, spellings, and conversions for units of measure, the data in the yield column will never spread nicely across a table or provide pretty infographics.Footnote ⁶⁵ Indeed, its confusing contents hammer home hardest of all the dissonance between the uniformity desired by the científicos and variation insisted on by rural producers. Placed in the context of decades of data collection, the yield column provides a window into how assumptions of consolidated capacity and their accompanying rhetoric undermined a shared attempt at information gathering.

The column as conceived imagined an orderly march of monocrop properties with consistent outputs that could be predicted in advance. The space provided allowed only a single rate of yield—and no indication that those providing information ought to note what was being grown. In the exchanges between the Department of Fomento and the Agricultural Society over 1898, as they hammered out the survey form, we can see the gradual elimination of columns asking for the names of crops. The elimination was based on the assumed capacity of Fomento officials to derive this information by cross-referencing the 1899 survey data with the piles of data the Office of Statistics already had on hand.Footnote ⁶⁶ Municipal officials filling out the surveys likely made such assumptions too, knowing that Mexico City officials must have additional information at their disposal and thus providing only the single data point the survey columns and rows suggested. Reports for about two-thirds of the properties included show only a single rate per unit and, perhaps, a metric, generally hectoliters or kilograms or seed employed. The metric might read something like 100 x 1, 50 for 1, or 20/1, with no indication of the crop to which this rate referred. Once again, the historian, separated by more than a century, finds herself sharing the contemporary bureaucrats’ dilemma: what to do with numbers without content? Yet, she also finds herself sympathetic to the producers and municipal officials. Why provide more information than was asked for, especially given the piles of surveys detailing crops they had previously completed?

A look at weather, though it is not a descriptor of property, does provide contrast and a reminder about context. Municipal officials used more than 30 different descriptors to complete the column titled clima (climate). In contrast to other columns, it is relatively straightforward to consolidate these descriptors into eight categories: cold (2,268 properties), temperate (7,508 properties), benign or healthy (452 properties), warm (2,631 properties), hot (1,130 properties), and varied (273 properties), plus categories for the 52 properties whose climate was represented with numerical data on temperature and the additional 65 that gave a weather descriptor outside these eight categories. Half the municipalities included only a single descriptor for their whole territory, often scrawled vertically across the whole column. In fact, some municipalities were spread across quite varied terrain, but this was a column that nonetheless invited and achieved an apparent consistency (see Figure 5).

Figure 5 Most prevalent standardized climate descriptor by municipality, mapped over North American Climate Zones. Prepared by author, 2023. Interactive map available https://caseylurtz.com/agricultural-statistics.

Map this data onto a contemporary depiction of Mexico's climate zones, though, and the localized referentiality of municipal officials’ terms comes into sharp relief. “Temperate” is the most common descriptor in the dataset by far, comprising more than half the properties reported. According to modern climate modeling, Mexico includes at least 13 different climate classifications on the basis of temperature, precipitation, and seasonality.Footnote ⁶⁷ An analysis of the use of the term “temperate” in the 1899 dataset reveals it to be spread across most of those zones. It is hard to believe that the person who described their home in Mexico's central plateau as temperate would use the same term to describe the weather in Chihuahua's northern deserts. Even within regions, officials clearly had different definitions of these terms. The Yucatán Peninsula, for example, aside from its northern coast, sits within a single modern climate zone—tropical savannah. Yet municipal officials there used a great variety of terms— bueno, benigno, regular, cálido, sano, templado, mal sano, frío, torrido, fresco (good, benign, regular, hot, healthy, temperate, unhealthy, cold, torrid, fresh)—to describe their home places.Footnote ⁶⁸ Even where the dataset seems to offer easy comparisons, it is an invitation undercut by reminders of referentiality and local variation.

Along with the municipal, state, and national cadastral surveys that were conducted for decades before 1899, the central government had also been collecting data on climate and crops. The referentiality with regard to weather that ArcGIS makes apparent to us today would also have been apparent to científicos at both Fomento and the Agricultural Society, after years of badgering local officials to report accurately on any crop in production, however miniscule the yield.Footnote ⁶⁹ Provincial officials and producers persisted in submitting such information as their names for different kinds of rain and the frequency of double rainbows, even as technocrats urged them to adopt emerging global norms for describing temperature, precipitation, barometric pressure, and so on.Footnote ⁷⁰ Similarly, local officials undermined assumptions of export-oriented cultivation by responding to questionnaires about crops in production with narratives of local consumption and intermixed subsistence and commodity crops.Footnote ⁷¹ Unable to subsume the responses into the normalized tables and figures it would like to have had, the Department of Fomento reprinted the transcript of local responses in full in its monthly bulletin.

Given that they had long accommodated the narrative and nuanced accounting of agricultural cultivation and weather, Fomento officials and Agricultural Society members alike should have known that providing a single space for yield information would result in a confusion of responses. And yet, the 1899 table included only a small, squished space for yield, barely enough for a single entry. Even before the editorial process took a hatchet to numerous columns, the draft version crafted by the Agricultural Society's farmers and agronomists left little space for reporting on multiple crops, despite their prevalence in rural life. Noting the limited space and assuming that the central government would apply the results of their previous survey responses, most municipal officials listed only one number or rate in the yield column. Even if I were to attempt to do the matching work that Fomento and the Agricultural Society never got to, the singular yield rate would prove hard to match to any one of the many crops each municipality reported elsewhere. The divergence in unspecified yields within many municipalities means that we cannot even assume that municipal officials were always reporting on the same crop within their jurisdiction. In a typical case, the municipal secretary in Nombre de Dios, Durango reported yield as “[number] por 1,” with those numbers ranging from 30 to 100.Footnote ⁷²

Turning to the almost 4,000 properties for which municipal officials wrote notes outside the table lines, there is no doubt as to the persistence of mixed cultivation of subsistence and commercial crops. Using multiple rows to describe yield for a single property, writing in letters as small as possible at odd angles to the sheet, or describing production for an entire municipality in a note below the table, municipal officials throughout the country reminded Mexico City technocrats of previous surveys answered.

Corn, always Mexico's staple good, was the most frequently named crop, whether listed alone or alongside other goods. Two-thirds of properties that only listed one crop listed corn, a crop that Fomento and the Agricultural Society had done little to study or promote to this point and whose inclusion was likely of little interest to World's Fair goers.Footnote ⁷³ Beans and wheat came next in popularity, followed by sugar, cacao, rice, and coffee. Of these, only cacao and coffee fall under our traditional understanding of export goods, though if we look to Mexico's export-import reports, we see that sugar, rice, and beans were also regularly sent abroad from the country's ports.Footnote ⁷⁴ Municipal officials resisted the national emphasis on export commodities and other tax-generating goods to continue reporting on corn and beans, even if other vegetables tended to get short shrift.Footnote ⁷⁵ It is tempting to speculate that the singular yield rate listed everywhere else corresponded to one of these crops, most likely corn, but again, the rates recorded are so varied, even within municipalities, that such a supposition would push the bounds of probability.

With some additional cleaning and standardization of metrics, we could generate comparisons of yield rates for those places that did specify both a crop and a rate, map them onto climate zones or proximity of transportation or market towns, or compare them with current-day production, or standard of living, or industrialization factors. To what end, though? The surveys were intended as advertisements of Mexico's current productivity and its potential. The diversity of crops noted by this subset of municipal officials does point to the country's embrace of a varied market basket. Further statistical or cartographic analysis could show which parts of the country presented themselves as most fruitful, most full of possibility. Chiapas, where many officials carefully noted yields for different export crops, was a hub for foreign and national investment in tropical agriculture in the late nineteenth century. It is not surprising that its officials touted the cacao, coffee, and sugar they produced. In contrast, Oaxaca and Yucatán also saw significant investment in agricultural modernization during this era and yet neither reported much in the way of yield data. As with so much of this dataset, further attempts at quantification are misleading at best.

Instead, the yield data reminds us of the shared and ongoing project of promotion and knowledge-gathering and the ways it was undermined in the course of its adoption and implementation. Municipal officials and Mexico City científicos alike assumed a degree of statistical and administrative capacity on the part of the central state. They assumed, for the most part, a shared language of crops and metrics based on decades of corresponding about agricultural productivity. These assumptions mean that the yield data provided by survey participants was and is unusable in the aggregate. It can provide insight for scholars working on a local scale who are interested in particular municipalities, so long as they already have a sense of what was growing and how to integrate the reported rates with regional market information. Yet overall, the yield column highlights the ways in which the state was blinded to the everyday norms of subsistence production by its own emphasis on exports.

The Means of Cultivation

What counts as a tool? A machine? An apparatus? Whose work is considered productive? The right-hand half of the table requested information on the means of production, both human and mechanical. Reporting on labor resulted in relatively consistent norms that nonetheless trouble our assumptions about wages across Mexico, while the columns on tools and machinery show both the penetration of technocratic understandings of what counted as agricultural modernization and rural people's continued reliance on much older aids in cultivation. Municipal officials in their majority pushed back on the insistence on new technologies, but by regularly making the distinction between old and modern, Mexican and imported, they made clear their engagement with ideas of agricultural progress. By reporting local labor norms almost exclusively in terms of daily wages, they demonstrated an embrace of liberal imaginings of free wage labor, though questions of gender, age, and the occasional slippage around that daily wage illuminate the cracks in the facade.

As with yield and climate, it makes sense to look at tools, machinery, and labor together in order to see the ways that both apparent standardization and incommensurability point to provincial reworkings and frustrations of the assumptions made by state projects. Technocrats were more specific in their requests for information in these columns than in those columns that came before. The categories “máquinas,” “áparatos e instrumentos,” and “promedio de los jornales diarios que se pagan cada hacienda o rancho” (“machines,” “tools and instruments,” “average of the daily wage paid on each hacienda or rancho”) were further divided. For “machines,” the information requested was number, manufacturer, and horsepower; for “tools and instruments,” it was number, name, and value; and for “wages,” the breakdown was for men, women, boys, and girls. Here was the specificity and definition of terms lacking in other parts of the survey. Only machines with a manufacturer mattered. Only tools with value needed to be reported. Only daily wage labor counted. Again though, municipal officials’ responses undermined the imagined consistency of these terms by chipping away at their implied norms.

The Department of Fomento and the Agricultural Society dedicated considerable resources to promoting the adoption of agricultural technologies by Mexican farmers. As with irrigation, mechanization and the use of sophisticated tools stood in for the advancement of rural productivity and the embrace of modern agriculture. As Alejandro Tortolero Villaseñor's careful study of innovation and agriculture in late nineteenth century Mexico makes clear, elites both in and beyond government pushed the country to move “de la coa a la máquina del vapor”—from the traditional digging stick to the steam-powered machine.Footnote ⁷⁶ This push was directed at both smallholders and well- capitalized enterprises, though of course it was most often the latter who had the means to acquire and put into use tractors, threshers, sorters, and mills. Edward Beatty's work on the history of technological importation and uptake further clarifies this by demonstrating how, while invention was part of the story and patents for new kinds of tools were encouraged, the definition of progress in agriculture and in industry tended to rely on adoption rather than innovation. Yet, even where education in new technologies was encouraged and facilitated by Fomento and other government entities, human capital of the sort represented by technological know-how remained scarce.Footnote ⁷⁷

The impulse for adoption and importation is quite clear in the dataset. So is the continued use of traditional implements. Though space in the columns was tight, municipal officials listed a multitude of tools employed, spilling their words into neighboring columns or referring readers to notes they added below the table when they ran out of space, or when the notes were providing information applied to their municipality as a whole. Beyond the generalizations of “varios” (various), coas, hatchets, and machetes were the most common implements listed, and more than half of all properties that enumerated their machinery included some sort of plow among their inventories.

Here the embrace of modernizing agriculture as represented by manufactured implements becomes apparent. Adjectives were common in these columns. They indicated whether tools were new or old, imported, or outmoded. Some plows were wooden, some were iron. Some plows were US-made, some were Mexican-made or “del país,” or common. Some were just old. Value was generally indicated per tool by kind, and here it was clear that the old-fashioned Mexican implements were valued less than the new US-made ones.

While US-made plows were strongly concentrated in the northern states of Tamaulipas and Coahuila, imported machinery was in use on varied crops throughout the country, from the coffee of Chiapas to the grains and beans of Guanajuato and the sugar of Jalisco. While correct spelling often eluded them (“Marshall,” of the British Marshall agricultural manufacturing enterprise, was spelled alternately as Marsai, Marsal, Marxal, Marzal, Marschall, and otherwise), officials took care in noting in as much detail as possible the make and model of the threshers, mills (molinos), sugar mills (trapiches), and shellers (desgranadoras) employed by larger properties or by the town as a whole. In the latter case, annotations made clear that not each property had a given machine, but rather that all had access to it, without clear indication of ownership.Footnote ⁷⁸ When it came to reporting fuerza motriz, or horsepower, responses highlighted how the survey's goals of uniformity again frustrated local understandings of agricultural progress and productivity. Many machines were still powered by actual horses or mules, rather than coal or steam, and water, a mano or de mano (by hand), and even sangre (blood) came into play.Footnote ⁷⁹ Some simply listed a rate of force, but in providing the kind of detail that mattered to them in the actual employ of tools and machinery, municipal officials kept this column from yielding standardized data, despite the carefully chosen column headers.

The request for daily wage by age and gender resulted in some of the most consistent and coherent data reported in the tables, a seeming conformity to the kinds of labor imagined and legislated from Mexico City (see Figure 6). Almost all officials included a daily pay rate in the column for men, often generalized for entire municipalities via repetition, ditto marks, or an annotation running down the column or below it. Fewer officials filled in the columns for women, boys, and girls, but there too the norm was a single number of cents per day. The realities of indebted labor, indentured labor, seasonal labor, familial labor, and imprisoned labor all faded in the face of regimented rows of individuals working at a set rate by the day.

Figure 6 Average Daily Wage for Men (in Pesos), Summarized by Municipality. Prepared by author, 2023. Interactive map available https://caseylurtz.com/agricultural-statistics.

Taking these rates at their face value, we see hired hands as commonplace no matter the size of an agricultural property; we see also the broad use of men, women, boys, and girls as laborers earning pay in their own right. Within municipalities, daily wages for men generally differed by a few cents between properties, with small and large alike listing rates of pay of between 2 cents and 1 peso. There was little correlation between the size of the property and the pay rate.Footnote ⁸⁰ Across the country, the average men's wage was 30 cents, the median 25. About 25 municipalities listed pay ranges, rather than specific rates, that revealed large internal disparities. Properties in Purificación, Jalisco, for example, listed ranges as great as 2 to 31 cents.Footnote ⁸¹ Municipal officials were slightly more likely to use the ditto marks or an annotation for the whole municipality to aggregate information for small properties than for large, but plenty of large haciendas, which respondents generally listed first, simply set the mark for the rest of the municipality.

Regionally, the agricultural survey casts doubt on our long-held generalizations about pay. Northern states consistently reported wages at the higher end of the range for male workers, but so too did export-oriented regions in the south. Historiography there has long focused on southern debt peonage and extortion, “slavery disguised” as the muckraking journalists of the era would have it.Footnote ⁸² While southern states like Oaxaca, Chiapas, Yucatán, Guerrero, and Morelos had some of the lowest daily wages reported nationally, they also had some of the highest in regions dedicated to crops like coffee, henequen, and sugar. Many of these municipalities also reported more and smaller properties than the average, undergirding work I have done elsewhere to show how the continuing availability of land in southern Chiapas undermined efforts to capture labor for plantation work.Footnote ⁸³ Municipal officials’ reporting in the 1899 agricultural survey suggests that similar outcomes may have been the case throughout the export-oriented south.

Officials reported less frequently on wages for women, boys, and girls than they did for men, and always reported them paid at lower rates.Footnote ⁸⁴ Women tended to make 50 to 75 percent of what men made (18 cents on average, a median of 15.5 cents). Boys consistently earned half of a man's wage (15 cents on average, with a median of 12 cents). Girls received a wage equivalent to a quarter to a half of a man's daily pay (11 cents on average, with a median of 12 cents). While municipal officials did not always include workers other than adult men in their tables, the presence of women and children as wage workers in the 1899 survey is stronger than national censuses from the years surrounding it would suggest. Representations of labor in those censuses were notoriously problematic, as the categories for kinds of work were drawn from the French Bertillon system which emphasized urban and industrial work over agricultural labor. As scholars such as Francie Chassen-López have shown, the flattening of rural work into a bare four columns, as compared to dozens for telegraphists and watchmakers and florists and so on, particularly underserved women's work. Her investigation of local reports on wages and property claims through repartition of village lands in Oaxaca makes clear that women represented a key part of rural life beyond the home.Footnote ⁸⁵ In the 1899 agricultural survey, even if the kind of work beyond day labor was not indicated, Fomento and Agricultural Society technocrats did strive to see and count the work everyone did.

Yet not every municipal official followed suit. An equal number of officials included notes on women's work as explicitly stated that women and girls were not employed, providing notes with explanatory phrases like “no hay” (none), “no se les da ocupación” (they are not given work), “no se ocupan” (they are not occupied [in this work]), or “no se ocupan en jornales” (they do not work for a daily wage). Other responses were “no se ocupan en la agricultura” (they do not do farm work), “casi no se ocupan las mujeres . . . por jornal” (women are generally not employed at a daily wage), “no se usa” ([they are] not used), “no se utiliza” ([they are] not utilized), and “no trabajan” (they do not work). These explanations also appeared in the column for boys, but less regularly than in the columns for women and girls: some 1,228 properties included phrases like this for women, 1,300 for girls, and only 288 for boys. Such language was used for men on only 19 properties.

Most municipal officials simply limited themselves to the survey item as given—the “promedio de los jornales diarios” (average of daily wage)—tailoring their local reporting the government's generalized push for contract wage labor represented in labor regulations and agricultural modernization literature of the day. Yet a few included language that made clear that other modes of employment persisted. The variety of phrases used to explain the absence of women's and girls’ wages is one place this can be seen, particularly in those phrases that made clear that it was the “jornal” part of the column—the daily wage—that resulted in women and girls’ exclusion. Thus, they implied that while women and girls might indeed take up agricultural work, they were remunerated for it in other ways, perhaps by having their labor covered by their husband's or father's wage, or through a sharecropping arrangement.

A very few municipalities further explained their local labor arrangements to show how the wording of the column did not do them justice. Some places listed monthly pay rather than the jornal diario, pointing to something more akin to a salary. Others instead specified the number of men working on the property and gave their total pay, bringing out the importance of having sufficient people available to labor, as well as highlighting a seemingly low average wage.Footnote ⁸⁶ One municipal official in Michoacán wrote that “The daily wage is very rare here because, in general, every farmer sows what he personally needs, in such small scale that the crop to which he dedicates himself is not worth mentioning.”Footnote ⁸⁷ Finally, three municipalities in Chiapas distinguished between wages for indebted laborers and wages for those not indebted, with indebted laborers earning a few cents less on average.Footnote ⁸⁸

The respondents’ insistence on differentiating among workers and their wages instead of providing a single answer points again to how the survey categories dictated by officials at Fomento and the Agricultural Society flattened local variation. The municipalities that insisted on providing large ranges for their daily wages instead of a single average figure, and the municipalities that distinguished between indebted and non-indebted workers (“el trabajador no es endeudado”)—and others—remind interpreters of the data that agricultural labor was not a uniform category of work. Beyond this, historians also remind us that the jornal was not always precisely daily rate of pay, but might instead be a piecework metric—based on the quantity of coffee picked or land tilled or sugar planted.Footnote ⁸⁹ Yet fewer municipal officials pushed back against the strictures of categorization with regard to labor than to other aspects of agriculture presented in the survey. Was this just exhaustion at the end of an overly detailed table? Was it a generalized embrace of wage labor as the desired mode of employment? Or was it a representation of the realities of increasingly formalized agricultural labor? Additional research based on hacienda records can bring us closer to answering these questions in years to come.