the use of spanish historical archives to … · viceroyalties were established: one in mexico...

1025AUGUST 2003AMERICAN METEOROLOGICAL SOCIETY |

T he use of documentary sources is a well-establishedtechnique for reconstructing climate during thepast few centuries, when no instrumental or

similar sources of data are available. Such sourceshave been used to study climate phenomena suchas El Niño–Southern Oscillation (ENSO; Quinn et al.1987; Quinn and Neal 1992; Ortlieb 2000) and theNorth Atlantic Oscillation (NAO; Jones et al. 1997;Luterbacher et al. 1999). They have also been usedto describe regional climate variability, especially inEurope and Asia. In recent years, substantial effortshave been made to obtain climatic information fromSpanish colonial archives covering the period 1500–

1900. Spaniards arrived in the Antilles at the end ofthe fifteenth century and, after 50 yr of constantadvance, they had established themselves throughoutmost of the southern part of the New World.Interestingly, southern South America, the most in-accessible region, provides some of the earliest his-torical records, such as Ferdinand Magellan’s navi-gation in 1520 of the strait that bears his name. Ascolonization advanced, new lands were incorporatedand new cities were founded. Valuable historical in-formation started to emerge from cities such as SantaMarta (1527) and Caragena de Indias (1553), in Co-lumbia; Lima, Perú (1532); Santiago de Chile (1541);Asunción del Paraguay (1541); and Mendoza, Salta,Jujuy, Tucumán, and Buenos Aires, all in Argentina(1561–80) (Morales Padrón 1973). Figure 1 showsthe location and foundation dates of the most impor-tant Spanish colonial cities.

The nature and origin of documents that can beused to reconstruct climate are extremely diverse.Some, such as ship logbooks, city council records, andchronicles, originated from reporting requirementsimposed by the Spanish Crown; others, like travelers’sdescriptions and diaries, resulted from the initiativeof private citizens. The reports of scientific expedi-tions are also a valuable source of climatic and envi-ronmental information, as are the large number of

THE USE OF SPANISH HISTORICALARCHIVES TO RECONSTRUCT

CLIMATE VARIABILITYBY RICARDO GARCÍA HERRERA, ROLANDO R. GARCÍA, M. ROSARIO PRIETO,

EMILIANO HERNÁNDEZ, LUIS GIMENO, AND HENRY F. DÍAZ

Spanish historical archives are a potentially important source of climate-related information

over a wide range of timescales and geographical locations.

AFFILIATIONS: GARCÍA HERRERA, AND HERNÁNDEZ—UniversidadComplutense, Madrid, Spain; GARCÍA—National Center forAtmospheric Research, Boulder, Colorado; PRIETO—CRICYT,Mendoza, Argentina; GIMENO—Universidad de Vigo, Orense, Spain;DÍAZ—Climate Diagnostics Center, NOAA, Boulder, ColoradoCORRESPONDING AUTHOR: Rolando R. García, National Centerfor Atmospheric Research, 1850 Table Mesa Drive, Boulder, CO80303E-mail: [email protected]: 10.1175/BAMS-84-8-1025

In final form 2 May 2003©2003 American Meteorological Society

1026 AUGUST 2003|

reports written by members of religious orders (see,e.g., Udías 1996). However, not all of these documentsare equally useful as sources of climatic information.Thus, the reports from expeditions or travelers canbe very detailed, but they usually refer to very shortperiods of time, so they are not particularly well suitedfor studying climate variability. On the other hand,local chronicles, although usually not very detailed,span much longer periods and can yield an adequateperspective on variability over interannual, decadal,and even longer scales.

These documents are presently held in differentarchives, either in Spain or Spanish America. In Spain,the most significant are the Archivo General de Indias(General Archive of the Indies; AGI), in Seville, theArchivo del Museo Naval (Archive of the Naval Mu-seum; AMN), in Madrid, the Archivo Central de laMarina (General Archive of the Navy; AGM), in Visodel Marqués, the Archivo General de Simancas (Gen-eral Archive of Simancas; AGS), in Simancas and theArchivo Histórico Nacional (National HistoricArchive; AHN), in Madrid. The AGI, in particular,

contains a vast amount of information covering allaspects of Spain’s relations with its American colonies,while the AMN and AGM hold extensive collectionsof logbooks.

The contents of the archives are usually dividedinto “sections” or “series,” each one covering a dif-ferent area of administrative, social, political, or mili-tary affairs. Within each such division are found ahighly variable number of legajos, or manuscriptbundles, consisting of some 1500–2000 manuscriptpages on related topics. The organization of the ar-chives varies greatly, depending on their origin andpurposes. For example, the organization of the AGIreflects the administrative structure of the Spanishcolonies more than any other criterion (see, Garcíaet al. 2001, for a more detailed description on theAGI). On the other hand, the AMN is organizedaccording to the origin of its collections, and com-prises four main divisions: Depósito Hidrográfico(Hydrographic Depository), Real Compañía deGuardamarinas (Royal Company of Navy Cadets),Fondos de Adquisiciones y Donaciones (Acquisitionsand Donations), and Museo Naval (Naval Museum).

Other significant documents have survived inSpanish America itself; these are presently kept indifferent cities, in provincial or national archives.Depending on the administrative status of the city,the information preserved may refer to the region inthe immediate vicinity of the city, or to much widerareas under the city’s jurisdiction. The most impor-tant administrative and political unit in SpanishAmerica was the viceroyalty. Starting in 1544 twoviceroyalties were established: one in Mexico (NuevaEspaña), which included all the territories north ofpresent day Panama, plus Venezuela and the Carib-bean Islands, and a second one in Lima (Peru),governing the rest of the Spanish possessions. Duringthe eighteenth century two new viceroyalties werecarved out from Lima: Nueva Granada (1739), withthe capital in Santa Fe de Bogotá and jurisdiction overtoday’s Venezuela, Colombia, Panama, and Ecuador;and Río de la Plata (1776), whose capital was BuenosAires and governed Bolivia, Paraguay, Uruguay,Argentina, and part of Chile. The highly centralizedand bureaucratic character of the Spanish empire ledto the production of a huge volume of documents,which provide a thorough coverage of the colonialterritories. This administrative hierarchy has a coun-terpart in present-day archives. Thus, the NationalArchives at Mexico City and Lima provide the great-est wealth of information, usually extending beyondthe present borders of these countries. However,additional information from the cabildos (local city

FIG. 1. Map of the American continent showing the lo-cation and date of foundation of the most importantSpanish colonial cities, and the four principal subdivi-sions of Spain’s colonial empire (viceroyalties of NuevaEspaña, Nueva Granada, Peru, and La Plata).


councils) can be found in a number of local and pro-vincial archives. The organization and structure ofthe cabildo archives varies, depending on the coun-try and the city.

The goal of this paper is to describe and discusssome strategies and techniques used to obtain climate-relevant information from these documentaryrecords. The next three sections illustrate, by meansof specific examples, three types of situations that areoften encountered when trying to extract usual datafrom historical archives: the treatment of noninstru-mental meteorological observations, the interpreta-tion of documents that provide explicit informationon climatic events or impacts, and the developmentof climate proxies. The final section summarizes theprincipal methodological issues and provides someguidelines, derived from the authors’ own experience,that may be useful for other researchers interested inobtaining climatological information from historicalsources.

NONINSTRUMENTAL OBSERVATIONS.Climatologists are most interested in long series ofobservations that can be used to define climatenormals and to study climatic variability. Spanishcolonial archives contain a certain amount of suchdata; perhaps the most obvious example being thelogbooks of merchant and military ships. A royal or-der issued in 1575 required masters and pilots ofSpanish ships in the Carrera de Indias (the route fromthe mainland to the American colonies) to keep arecord of each trans-Atlantic journey, including adetailed description of the voyage and of any geo-graphical discoveries, winds, current, and hurricanes.The completed logbooks had then to be delivered tothe Professor of Cosmology in the Casa deContratación (AGI, Indiferente General, 1956, L1,f 266r–266v).1 As the Spanish Administration evolved,the logbooks were kept by different authorities; at theend of the colonial period, the logs were archived byeither the Navy or the Post Office authorities.

By the beginning of the eighteenth century, withmany ships traveling between the metropolis and thecolonies, the logbooks include detailed records of at-mospheric and oceanic conditions. Toward the endof the eighteenth century, as barometers and ther-mometers came into common use, the earlier descrip-

tions of weather events began to be accompanied byinstrumental records. Nonetheless, even the earlierlogbooks, from the sixteenth and seventeenth century,contain much reliable information because theircrews included highly skilled officers, well aware thattheir lives and careers depended on the ability to ob-serve the weather (Wheeler 1988). Figure 2 shows areproduction of a page from the logbook of the frig-ate San Andrés, on her way from Manila, Philippines,to Acapulco, Mexico in 1796 (AMN, Ms 201).2 It pro-vides hourly data on the heading of the ship (rumbo),wind direction (viento), strength (fuerza), state of theatmosphere (caris, or cariz in modern Spanish), rig-ging (aparejo), and direction and distance run by theship. Midday position and a brief description of themost relevant daily events are entered at the bottomof the page.

Logbooks are a unique source of historical databecause their purpose is to describe, as precisely aspossible, the state of certain meteorological variables(wind strength, wind direction, precipitation, etc). Inprinciple the data they contain resemble those of theinstrumental period, except that descriptive terms areused instead of numerical values. Fortunately, thou-sands of these logbooks—mainly those kept by thefleets that sailed the Atlantic—have survived and areavailable in different archives. They are structuredalong the lines of the example shown in Fig. 2, andprovide high-resolution (sometimes hourly) observa-tions, with a high level of correlation among logbooksfrom different ships and officers.

Interpretation of meteorological entries fromships’ logs is one of the most immediate and straight-forward applications of historical data to climate stud-ies. However, it is not at all a trivial exercise, especiallywhen large amounts of data are involved, becauseextensive processing is necessary to convert the hand-written logbook entries into a computer database suit-able for quantitative study. Systematic transcriptionof logbook data from Spanish archives is currentlybeing carried out under the Climatological Databasefor the World’s Oceans 1750–1850 (CLIWOC)project. This project is funded by the EuropeanUnion, and its chief aim is to obtain some one mil-lion logbook observations from Spanish, British,Dutch, and French archives for the period 1750–1850(see www.ucm.es/info/cliwoc for a detailed descrip-

1 References to manuscript sources in the Archivo General de Indias are denoted by the initials AGI, followed by the name of thesection of the archive where the manuscript is located, and a series of numbers identifying the legajo, or bundle, to which themanuscript belongs.

2 Documents in the collection of the Museo Naval of Madrid are catalogued simply by the legajo number.

1028 AUGUST 2003|

tion). In what follows we de-scribe the processing requiredto obtain wind data compa-rable to that available frommodern records.

As can be seen in Fig. 2,wind observations refer toboth speed and direction. Theinterpretation of wind direc-tion records is straightfor-ward, because standard com-pass directions were used.Wind speed data, on the otherhand, require much morecareful work. Of course, noanemometers were availableon board these sailing ships,but, as previously mentioned,the officers were highly skilledin estimating wind from thestate of sea, sails, and clouds.These estimates were recordedusing descriptive terms ratherthan expressed numerically.In the case of Spanish log-books, a large variety of termswere used to classify the inten-sity of the wind; recent workby García Herrera et al. (1999)has identified as many as 36different terms used to de-scribe wind intensity.

Thus, the first task in thistype of study is to construct adictionary to translate the his-torical terms into modernequivalents; the use of con-temporary and historical ma-rine dictionaries is indis-pensable for this purpose.However, this is not enoughbecause nautical terminologyusually evolves significantlyover the time span of a cen-tury, so the dictionary itselfneeds to evolve to cope withthese changes. Content analy-sis (Baron 1982) can be a use-ful tool to overcome this kindof problem. Once a dictionaryhas been completed, the origi-nal readings can be convertedinto their modern equivalents

FIG. 2. Reproduction of a page from the logbook of the frigate San Andréson her way from Manila to Acapulco (AMN, Ms 201; reproduced with per-mission of the AMN, Madrid, Spain). The title at the top of the page showsthe period of record (from Friday, 5 to Saturday 6 Aug 1796). The mainsection contains entries for the time of day, heading, wind direction andstrength, state of the weather, rigging, etc (note that the 24-h day beginsat local noon, standard practice on Spanish ships in the eighteenth century).The summary at the foot of the page reads approximately as follows: “Webegan this leg under a Galeno [a category 4–5 wind on the Beaufort scale]and fair weather. [The wind then] began turning to the NW, the weatherworsening and becoming showery, and at 7 [p.m.] it turned to the N with somestrong gusts, at which time lightning started in the 1st, 4th, and 3rd, quadrants.The wind [then turned] from N to W, with strong gusts from the NW. A stormhaving formed WNW, at 11:30 [p.m.] the wind blew strongly from the NW withheavy rain and gusts. Shortly thereafter, the wind slackened, blowing from theW and SW, with lightning continuing on the NE. The weather [later] worsenedand, at 14 hs. [2 a.m.] the wind blew again from the W and NW, with loudthunder, lowering clouds and heavy rain. The wind [then] turned suddenly tothe NE with strong gusts until 3 [a.m.], when it started to weaken, the rainceased, and the horizon became clearer. At dawn, it rained until 6:30 [a.m.],when the wind turned to the SW.”


and introduced into the database. In spite of these dif-ficulties, intercomparison exercises (e.g., Wheeler1988) demonstrate the coherence of the informationfrom logbooks which can, therefore, be consideredhighly reliable.

Similar procedures must be repeated for othervariables of interest that are often found in logbooks,such as precipitation, cur-rent weather, or the state ofthe sea. Only then can thedata be considered to beavailable for use by the sci-entific community. Clearly,successful extraction ofweather data from ships’logbooks requires a major,coordinated effort using acombination of expertisefrom fields such as clima-tology, history, and linguis-tics, among others. Figure 3shows the average wind di-rections for the postal routefrom La Coruña, Spain, toHavana, Cuba, back to LaCoruña, from a sample of166 trips for the period1764–69. A 5° × 5° grid hasbeen used to represent winddirection. It is expected that, when all the logbooksare finally examined, the database will provide evi-dence on the variability of the Azores subtropicalhigh.

DESCRIPTIONS OF WEATHER EVENTS.Explicit meteorological data, such as that describedabove, will be immediately familiar and useful to cli-matologists. However, the historical archives containother types of information whose application is notas immediate as that of the logbook data. This is usu-ally in the form of reports about specific weatherevents or impacts in a certain city or region (e.g., cropreductions, destruction of buildings). The use of suchsources is complicated by the fact that few documentsdeal specifically with weather; instead, meteorologi-cal information has to be gleaned from documents ofa very diverse nature. An outstanding example are theActas Capitulares, or city council records.

The Actas have been especially fruitful for climaticstudies. They summarize the meetings of the cabildos.Although the cabildos were a colonial institution,they often remained in existence well into the repub-lican period in many of the former Spanish colonies.

The cabildos met weekly, usually on Saturdays, andevery session was carefully documented by theescribano (recorder) who had the responsibility ofproducing a faithful account of the meeting and car-ing for the books. The members of the cabildo werein charge of many aspects of the affairs of the city: itsmilitary and defensive organization, water supply,

economy, etc. In addition,they were invested with thepower of issuing local laws.Their role was especiallyimportant when the cityfaced some crisis, such as aflood, drought, or earth-quake, that impacted thelives or livelihood of thecitizens. The Actas areamong the most reliablesources of climatic infor-mation partly because ac-counts of weather events ordescriptions of the climateare not their main purpose.Thus, the chances of inten-tional distortions of theevents described are lessthan with other documents.Additionally, when collec-tions are well preserved,

the Actas constitute a continuous source with dailyor weekly resolution, spanning long periods.

Unfortunately, not all the Actas have survived.Wars, political instability, and poor archival condi-tions, especially during the unstable nineteenth cen-tury, have resulted in the loss of some collections.However, in many cases they have been adequatelypreserved and have even been published. Someexamples include the complete collections for theArgentine cities of Santiago del Estero (covering theperiod 1554–1810; Academia Nacional de la Historia1941–1951), and Corrientes (Academia Nacionalde la Historia 1941–1946). From La Paz (Bolivia),only those for the period 1548–52 have survived. Onthe other hand, Potosí (Bolivia) has preserved a col-lection of exceptional quality that has been studiedrecently by Gioda and Prieto (1999). The extractionof information from the Actas requires extensiveknowledge of the contents and organization of thearchive in question. Extreme events, such as floods,droughts, or hurricanes, are especially well recordedin these documents, as is shown in the followingexample:

FIG. 3. Reconstruction of the average winddirections using logbook data for the postalroute from La Coruña to Havana to LaCoruña, from a sample of 166 trips in theperiod 1764–69.

1030 AUGUST 2003|

Actas Capitulares de laprovincia de Mendoza (Ar-gentina) 11 de abril de 1662(ACM, Vol. 3, p. 220)3

“. . . por cuanto martes amedianoche 11 del corrientefue Dios nuestro Señorservido de enviar unaavenida conque asoló yarruinó la mayor parte dela ciudad y entre ellasfueron las casas del Cabildoy cárcel que no se puedenhabitar ni hacer juicio enellas por estar caidas . . .”[. . . whereby at midnighton Tuesday the 11th of thecurrent month, it was thewill of our Lord to send aflood that leveled and ru-ined the greater part of thecity, including the house ofthe cabildo and the jail, sothat they have become un-inhabitable and unsuitablefor holding trials since theyhave collapsed . . .].

Drought in northern Argentina.An example of the use of theActas and similar historicalsources to detect climatefluctuations can be found inPrieto et al. (2000). This paperexamines climate variability inseven locations in northernArgentina during the six-teenth–nineteenth centuriesand illustrates the need for integrating informationfrom different archives and types of document. Thestudy employed Actas Capitulares from the area(found in local archives and in the AGI); local reportson plagues from local archives; miscellaneous admin-istrative documents from the AGI; RelacionesSemestrales (semestral reports on precipitation andcrops produced under an order of the Crown dur-ing the eighteenth century, which are archived locallyand in the AGI); diaries from private citizens, trav-

elers, and members of religious orders; and recordsfrom missions (found in local archives, in the AGI,or published separately).

From these sources, continuous series of annualanomalies of precipitation, runoff, and floods for theselected area for the period 1580–1806 were devel-oped (see Table 1 of Prieto et al. 2000). Due to theheterogeneous character of the primary sources, care-ful interpretation and cross-checking were requiredto obtain a homogeneous record before content analy-sis could be applied to convert the original informa-tion into a qualitative, homogeneous scale. In this way,anomalies were classified within five categories forprecipitation (very wet, wet, normal, dry, and verydry) and within two categories for the occurrence of

3 The Actas Capitulares of the province of Mendoza are orga-nized according to bound (manuscript) volumes within theprovincial archive; however, they have not been published.

FIG. 4. Decadal variability of climatic conditions in northern Argentina dur-ing the eighteenth century. The letters on the maps denote indices for pre-cipitation (W: very wet, w: wet, n: normal, d: dry, D: very dry) and flooding(f: some floods, F: massive floods). For every decade, the two most frequentprecipitation conditions have been indicated. The first panel shows the lo-cations used to characterize the climate of the region (1: Northwestern, 2:Arid Chaco, 3: Humid Chaco, 4: Cordoba humid plains, 5: Mendoza aridand semiarid plains, and 6: Rio de la Plata basin). Conditions for the period1970–79 have been included in the last panel for comparison. Reproducedfrom Prieto et al. (2000).


floods (some floods and massive floods). Althoughthis procedure is time consuming and requires a highlevel of expertise in the interpretation of the termi-nology employed in the original documents, the useof primary sources and overlapping references to thesame phenomenon makes for a high degree of reli-ability. Figure 4 shows the decadal variability of cli-matic conditions in northern and central Argentinaduring the eighteenth century. For each location, theprevailing conditions for every decade have been in-dicated. It can be seen that the period 1780–1800 isone of frequent, severe droughts and high precipita-tion variability compared to earlier decades of theeighteenth century and to modern values (1970–79,shown in the lower right-hand panel). This demon-strates how the combined use of the documentarysources provides a century-scale perspective on theregional climate of the area that cannot be obtainedfrom other sources.

Atlantic hurricanes. Another example of the use ofweather descriptions in climate studies is the recon-struction of Atlantic hurricanes (García Herrera et al.2002). This paper describes the results of anonexhaustive search in different Spanish archives(see the introduction for a list). It shows that the ar-chives contain abundant references to weather disas-ters on the Caribbean coasts. They usually refer tohurricanes or intense storms and describe damages inlives and property. For Cuba and Puerto Rico, in par-ticular, there are records covering more than 400 yr(approximately 1500–1898). Of special interest are theReports of the General Navy Command of the Antilles,the Gazetas, and other periodic publications, such asthe Partes de tranquilidad (included in the Papeles deCuba section of the AGI), the monthly papers pro-duced by the authorities, or the reports from the me-teorological services established in the area during thenineteenth century.

The occurrence of hurricanes in the open oceanis better described in ships’ logs, especially those fromthe annual silver fleets that traveled between theSpanish main and the metropolis. In turn, the log-books can be complemented by other types of re-ports, for instance, claims for compensation for lossesof merchandise in shipwrecks. As shown by GarciaHerrera et al., 127 hurricanes or severe storms wereidentified using the information extracted from thesesources. Figure 5 provides a summary of the results.It can be seen that most of the information refers tothe second half of the eighteenth century. It is stillunclear whether this is the result of a higher fre-quency of hurricanes or greater availability of

sources. Regarding spatial coverage, Cuba, PuertoRico, Florida, and Louisiana are the areas with themost available information.

Figure 5 also illustrates one of the main problemsinvolved in this type of study: the need for a carefuland systematic search of the archives. Fourteen sec-

FIG. 5. Distribution of the hurricanes identified fromhistorical documents according to (a) archive and sec-tion, (b) geographical area, and (c) in 50-yr intervalsbetween 1500 and 1898.

1032 AUGUST 2003|

tions in three different archives were identified as con-taining references to hurricanes, in many cases withunexpected results. For instance, it was not expecteda priori that sections of a rather bureaucratic nature,such as Contratación (Contracts) or Patronato (RoyalPatronage), could provide a high percentage of thereferences to hurricanes. Thus, although the searchprocess can be very time consuming, it can yieldunique information. In this admittedly preliminaryexercise, about 30% of the hurricanes were unre-ported by previous authors. In fact, most of the pre-vious historical series of Atlantic hurricanes have beenconstructed from sources in the Americas, but thepeninsular archives had hitherto remained unex-plored for these purposes. This highlights the valueof these sources, because they can provide indepen-dent evidence to refine hurricane chronologies,tracks, and effects.

Reevaluation of historical chronologies of ENSO. A finalexample illustrates the pitfalls associated with the in-terpretation of documents that describe meteorologi-cally significant events. Because the sources, authors,and purposes of these documents can be varied, dif-ferent aspects need to be considered when assessingtheir reliability. For instance, it is important to ascer-tain whether the sources are primary (original docu-ments, with firsthand information) or secondary(studies or inferences made from other documents,or relations from people who were not actual wit-nesses to the facts). If the sources are not original, dif-ferent interpretations can lead to disparate results; therecent revision by Ortlieb (2000) of previous chro-nologies of ENSO by Quinn and Neal (1992) andQuinn et al. (1987) is a case in point.

The papers of Quinn and coauthors are one of themost widely referenced chronologies of ENSO, butthey have been developed mostly from secondarysources, such as reports from floods, droughts, lossof crops, plagues, and migration of fisheries from dif-ferent locations in South America (mainly Peru,Bolivia, and Brazil). In addition, somewhat esotericindicators, such as flow volume for the Nile River inEgypt, have been included in the analysis on the ba-sis of possible teleconnection effects. Ortlieb (2000)critically reviews these results, evaluating the reliabil-ity and significance of the sources used to identify eachENSO event. As a result of this new analysis Ortliebconcludes that, of the 86 ENSO events identified byQuinn et al., 42 are doubtful. In addition, Ortliebidentifies seven events not previously associated withENSO by Quinn and coauthors.

INTERPRETATION OF CLIMATE PROXIES.Documentary sources have been used in the past asclimate proxies, particularly for the study of majorteleconnection patterns. The term “proxy” refers tovariables that are related to climate phenomena, butare not direct observations or descriptions of weatherevents. Historical sources can provide valuable prox-ies but, as already noted, historical data are often sub-ject to problems of interpretation. Most often thisoccurs because of shortcomings of the data itself; forexample, is the apparent variability in the frequencyof Atlantic hurricanes real or is it affected by non-uniform sampling? Are the proxies chosen to iden-tify ENSO events unequivocally related to theseevents? On the other hand, it is also possible to en-counter a proxy series wherein the data themselvesare unambiguous, but their geophysical significanceis unclear. The use of modern analogs can be a use-ful tool for interpreting the meaning of the series insuch cases, as done in the recent study by García et al.(2001).

This study focused on the climatic information thatcould be derived from the voyages of the Manila gal-leons. These ships carried out a regular commercialroute between the Philippine Islands, and Acapulco,Mexico, throughout the period 1565–1815. Becausethey operated regularly and the Manila-bound shipssailed in the tropical Pacific, near 12°N, documentsdescribing the voyages have obvious potential assources of climatic information. Unfortunately, andin contrast to the Atlantic voyages described in thesection titled “Descriptions of Weather Events,” veryfew logbooks from the Manila trade have survived.These were first and foremost commercial voyages,financed by private parties, and the route was so re-mote from the metropolis that many Crown regula-tions were disregarded. In any case, less than 10% ofthe Manila–Acapulco voyages are documented in log-books, so that the rich, high-resolution observationscommon in the records of the Atlantic fleets are notavailable.

Instead, it was possible to obtain the dates of de-parture and arrival at port from documents heldmostly by the AGI. Thus, a series of duration of thevoyages based on objective data (the dates recordedin the documents), and to all appearances free of sub-jective influences, was generated. A number of stud-ies were conducted to determine the potential influ-ence of changes in the starting date of the voyages,ship design, type of cargo, and route (see García et al.2001 for more details). All of them led to the sameconclusion: climate variability was the likely cause of


the behavior detected in the series. The problem thenbecame one of inferring the significance of the series,which showed very large secular trends during themid-seventeenth century, as can be seen in Fig. 6.

The duration of a sailing voyage between Acapulcoand the Philippines is an integral measure of thestrength and direction of the winds along the trajec-tory followed by the ships; however, it lacks specificinformation on any details of geophysical interest. Forexample, did secular trends in voyage duration reflectuniform changes in the trade wind circulation, or waslocal variability more important and, if so, where? Theuse of modern data provided the key to a plausiblesolution of this problem. García et al. (2001) con-structed “virtual voyages,” wherein the trajectories ofidealized ships sailing from Acapulco to the Philip-pines were computed using winds for the period1948–99 obtained from the National Centers forEnvironmental Prediction–National Center for At-mospheric Research (NCEP–NCAR) reanalysis(Kalnay et al. 1996; Kistler et al. 2001). The resultsshowed that the occurrence of the slowest or fastestvirtual voyages was closely related to the position ofthe southwest monsoon trough during the month ofJune, when the ships approached the Philippine ar-chipelago. In short, displacement of the monsoontrough north and eastward of its ensemble-mean po-sition resulted in slow virtual voyages, and viceversa. Figure 7 illustrates this result.

Thus the analysis of the virtual voyages led to thehypothesis that the secular trends observed in the his-torical series of voyage duration were likewise attrib-utable to changes in the southwest monsoon circula-

tion. The plausibility of this hypothesis was reinforcedby comparison with recent studies (e.g., Fu andFletcher 1988) that demonstrate that large, long-termchanges in the strength and position of the southwestmonsoon trough have occurred throughout the twen-tieth century. Taken together, these results provideda strong indication that the atmospheric circulationof the western Pacific underwent large, multidecadalfluctuations during the seventeenth century.

FIG. 6. The time series of voyage duration fromAcapulco, Mexico, to the Strait of San Bernardino, inthe Philippines, 1591–1750. The 30-yr running mean isdenoted by the solid line; the dashed lines indicate themean ±1 std dev.

FIG. 7. Mean Jun wind field (m s−1) for ensembles of 9 yrwith the fastest and slowest virtual voyages. Arrowsindicate the vector wind, and colors the wind magni-tude (m s−1). The north and eastward shift of the south-west monsoon between the fastest and slowest voyageyears is evident. The NCEP–NCAR reanalysis data wasprovided by the National Oceanic and AtmosphericAdministration–Cooperative Institute for Research inEnvironmental Sciences (NOAA–CIRES) Climate Di-agnostics Center.

1034 AUGUST 2003|

CONCLUDING REMARKS. The extraction ofclimatic information from documentary sources canbe a very time-consuming effort. In many cases thedocuments have not been scanned or catalogued elec-tronically, so that the originals have to be consulted.Even when documents are available in digital formand are correctly catalogued, little information can befound under meteorological or climatological head-ings. Thus, a search of many different sections of thearchives and diverse documents is almost always re-quired. Finally, the miscellaneous nature of the ar-chives often requires a high level of expertise in areassuch as history and paleography.

In the study of the Manila galleons just discussed,thousands of documents from eight different sectionsof the AGI sections were consulted. For every voyage,the following data had to be obtained to determine itsduration: the name of the ship, date of departure fromthe Manila, date of arrival at and departure fromAcapulco, and date of arrival at Manila. Similarly,thousands of pages of Actas Capitulares were read toobtain the raw information needed to establish theprecipitation anomalies documented by Prieto et al.(2000). In the exploratory work on Atlantic hurricanefrequency by García Herreraet al. (1999, 2001, 2002)hundreds of documents contained in the AGI, AHN,and AGS have been searched. Even when relevantdocuments are identified, great effort is often requiredto make them useful for climatological research. Thus,under the CLIWOC program, the development of thedatabase involves reading more than 600 original log-books, each of them containing hundreds of hand-written pages, transcribing the information into digi-tal form, translation of the entries, and final editingto make a friendly database for the end user.

When one examines the examples mentionedthroughout this paper from the methodological view-point, it can be seen that each type of project requiresits own approach, often developed “ad hoc.” Thus, forthe Manila galleons a model had to be derived to de-scribe the relationship between wind force and shipspeed, virtual voyages had to be designed, and mod-ern analogs constructed. For CLIWOC, it is necessaryto develop “dictionaries” of meteorological terms tomake the historical information, originally in fourdifferent languages, useful and comparable with mod-ern data. The characterization of climate variabilityin Argentina required extensive application of con-tent analysis. We have also seen that it is indispens-able to combine the expertise of historians, climatolo-gists, and meteorologists if it is desired to extractmaximum information from historical documents.None of the studies described herein would have been

possible had the research teams not combined suchabilities.

To summarize, in order to design an effective re-search strategy, it seems necessary to have the following:

• information from diverse archival sources, both inSpain and in the Americas;

• a high level of archival expertise to extract the rawdata, because the archives are usually organized ac-cording to administrative or historical consider-ations, and the use of geographical or climatologi-cal entries is not always sufficient;

• a preliminary historical–climatological analysis ofthe type of events and impacts for which data aresearched (which will usually produce a more effi-cient search and is preferable to blind searcheswithout any prior joint search criteria);

• a good grasp of the historical context to properlyanalyze the documents, assess their reliability, andobtain metadata that can be helpful inpostprocessing; and

• expertise in the analysis of meteorological and cli-matological instrumental observations, which willusually help in extracting the full significance ofnoninstrumental data.

These complexities and difficulties should not dis-courage researchers interested in the field, becausethe process can yield very valuable results in the longrun. The use of historical sources makes it possibleto study climate phenomena over extensive geo-graphical areas and on a broad range of timescales.And it must be borne in mind that, for many situa-tions of climatic interest spanning several centuriesbefore the instrumental era, there are simply nosources of information other than those contained inhistorical archives.

REFERENCESAcademia Nacional de la Historia, 1941–1946: Actas

Capitulares de Corrientes. 4 Vols., Kraft, 1650 pp.——, 1941–1951: Actas Capitulares de Santiago del

Estero. 6 Vols., Kraft, 2340 pp.Baron, W., 1982: The reconstruction of eighteenth cen-

tury temperature records through the use of contentanalysis. Climate Change, 4, 385–398.

Fu, C., and J. Fletcher, 1988: Large signals of climaticvariation over the ocean in the Asian monsoon re-gion. Adv. Atmos. Sci., 5, 389–404.

García, R. R., and Coauthors, 2001: Atmospheric circu-lation changes in the tropical Pacific inferred fromthe voyages of the Manila galleons in the sixteenth–


eighteenth centuries. Bull. Amer. Meteor. Soc., 82,2435–2456.

García Herrera, R., M. R. Prieto, L. Gimeno,E. Hernández, E. López, and R. Herrera, 1999: TheArchivo General de Indias as a source of climaticmarine surface information. Proc. Int. Workshop onDigitization and Preparation of Historical SurfaceMarine Data and Metadata, Toledo, Spain, WorldMeteorological Organization, WMO/TD 957, 61–68.

——, and Coauthors, 2001: El Galeón de Manila y elclima del Pacífico durante el siglo XVII. Estudios deHistoria y Ambiente en América, B. García Martínezand M. del Rosario Prieto, Eds., El Colegio deMéxico/IPGH, in press.

——, F. Rubio, D. Wheeler, E. Hernández, M. R. Prieto,and L. Gimeno, 2002: The use of Spanish and Brit-ish documentary sources in the investigation ofAtlantic hurricane incidence in historical times.Hurricanes: Past, Present and Future, R. Murnaneand K. Liu, Eds., Columbia University Press, in press.

Gioda, A., and M. R. Prieto, 1999: Histoire dessécheresses andines. Potosí, El Niño et le Petit AgeGlaciaire. La Météorologie, 8 (27), 33–42.

Jones, P. D., T. Jonsson, and D. Wheeler, 1997: Extensionof the North Atlantic Oscillation using early instru-mental pressure observations from Gibraltar andsouth-west Iceland. Int. J. Climatol., 17, 1433–1450.

Kalnay, E., and Coauthors, 1996: The NCEP/NCAR 40-Year Reanalysis Project. Bull. Amer. Meteor. Soc., 77,437–471.

Kistler, R., and Coauthors, 2001: The NCEP–NCAR50-year reanalysis: Monthly means CD-ROM anddocumentation. Bull. Amer. Meteor. Soc., 82, 247–268.

Luterbacher, J., C. Schmutz, D. Gyalistras, E. Xoplaki,and H. Wanner, 1999: Reconstruction of monthlyNAO and EU indices back to AD 1675. Geophys. Res.Lett., 26, 2745–2748.

Morales Padrón, F., 1973: Historia del Descubrimientoy Conquista de América. Editora Nacional, 717 pp.

Ortlieb, L., 2000: The documented historical record ofEl Niño events in Perú: An update of the Quinnrecord (Sixteen through Nineteen Centuries). El Niñoand the Southern Oscillation, H. F. Díaz and V.Markgraf, Eds., Cambridge University Press, 496 pp.

Prieto, M. R., R. Herrera, and P. Dussel, 2000: Archivalevidence for some aspects of historical climate vari-ability in Argentina and Bolivia during the 17th and18th centuries. Southern Hemisphere Paleo andNeoclimates, W. Volkheimer and P. Smolka, Eds.,Springer-Verlag, 57–69.

Quinn, W. H., and V. T. Neal, 1992: The historicalrecord of El Niño events. Climate Since A.D. 1500, R.Bradley and P. Jones, Eds., Routledge, 623–648.

——, ——, and S. E. Antúnez de Mayolo, 1987: El Niñooccurrences over the past four and a half centuries.J. Geophys. Res., 92, 14 449–14 461.

Udías, A., 1996: Jesuits’ contributions to meteorology.Bull. Amer. Meteor. Soc., 77, 2307–2316.

Wheeler, D. A., 1988: Sailing ship logs as weatherrecords: A test case. J. Meteor., 13, 122–126.