All of the data discussed below are subject to the caveat that a national bibliographic profile is being constructed as reflected in WorldCat. This means that it is subject to what libraries—and specifically libraries that participate in WorldCat—have collected and cataloged. For some countries in the present comparison, such as Thailand and Bolivia, very few libraries in each country have been OCLC members, and thus the data will tend to reflect more of what other Anglo-American libraries and other major national libraries participating in WorldCat have collected of their publications.

In addition, several issues of Anglo-American cataloging practice and other cataloging standards as translated into MARC 21 affect the following profiles. For example, different cataloging standards may have different (or changing) concepts of what will be coded as a book. Even more difficult for the purpose of this research is the definition of a country. The German data below, for example, reflect generations of different catalogers’ judgments of several centuries of shifting boundaries that have finally coalesced into what we in 2011 call Germany. The national bibliographic profiles of the current Balkan nations may be completely corrupted by the shifting assignments of MARC country codes in the region. Furthermore, cataloging practice in the area of date of publication may vary, especially with reprintings of prominent historical works. Finally, none of the historical profiles below should be taken as an assertion of historical causality on national book publication and collection.

The increasingly global and comprehensive nature of the WorldCat database should yield ample and relatively trustworthy representations of the six national bibliographic profiles being considered. In addition, the six basic historical publication profiles offer remarkably distinct and lucid images of each country's evolving publishing history and how books and library collections tend to reflect the history of each.

For the six case study countries, table 1 shows the scope of the basic dataset, with the ongoing caveat that these are the records represented in WorldCat. This gives an idea of the richness of the data possible to mine in WorldCat, country by country. German publications, not unsurprisingly, are the richest subset of these data; the records from Germany include, pertinently, the catalogs of the Deutsche Nationalbibliothek (DNB), the Bayerische Staatsbibliothek, and Hessische Bibliotheksinformationssystem (HeBIS), the consortium of the largest university libraries in the Federal Republic.²⁰ In addition, this country has spearheaded national efforts at bibliographic control under the leadership of the DNB.²¹ South Africa has a relatively high rate of holdings for its national publications, perhaps because of the participation of the University of Cape Town and the University of South Africa (Unisa) as cataloging partners in WorldCat.²² Of the six countries in this study, Bolivia has the fewest records in WorldCat; one can presume that their national bibliography is larger than 58,000 titles, but the lesser participation of Bolivian libraries in WorldCat to date somewhat weakens the publishing profile presented here.

Table 2 provides a comparative sample of several other countries’ datasets within WorldCat. It appears that for many countries (depending somewhat, but not entirely, on libraries’ contributions), the data mining will be able to profile a large set of publications throughout their history. Even in the case of a problematic country like Ukraine (a former Soviet republic), enough catalogers participating in WorldCat even during the Cold War have paid attention to specifying that Soviet publications from Kiev should receive the country code for the Ukrainian Soviet Socialist Republic (unr) rather than the blanket code for the Soviet Union (ur). Thus Ukraine's bibliographic heritage can remain separable from the Soviet Union in the bibliographic data. In the case of the national bibliography as seen in WorldCat for China, the data was mined in August of 2009 and will certainly offer a much more complete picture once bibliographic data from the National Library of China are fully integrated into the database.²³ The example of China highlights the importance of national libraries’ contribution to the aggregate bibliographic data available for data mining in projects of this type: whereas acquisition of a country's publications by American research libraries can offer the beginnings of a robust dataset, the bibliographic universe represented in WorldCat, Google, or any such data source is highly fortified by including the contents of national libraries.

A graphical representation of the data within a subset can at times be even more revealing. The robust data on book publication and library collection from Germany in the twentieth and twenty-first centuries (figure 1) reflect the historical dips in book publication from 1914 to 1919 and the complete collapse of the industry in 1945 and 1946. (No necessary historical correlation is being asserted, but rather the overlap of historical events and changes in the publication profile as shown in the data mining.) After German unification in the 1990s, German book publication itself may not have waned, but the merging of libraries and university systems within the now-unified nation may have led to fewer copies being held (thus being represented graphically in a parallel dip in both national and international holdings of German books). From a peak of more than 900,000 indigenous library copies of German publications in 1991 (the bulk in these data of almost 1.5 million copies worldwide), by the middle of the 1990s, the German libraries’ holdings of their own publications had dipped to around 780,000. The trough in “All Holdings” worldwide in figure 1 corresponds to the apparent culling of duplicate copies in German libraries.

The robust German dataset within WorldCat similarly documents some interesting overlaps with events earlier in the history of the German-speaking lands. In this earlier dataset (see figure 2), one can note several changes in recorded book publication that correspond to important historical developments. These include dips in both book printing and holdings during the Napoleonic Wars (1800s to the early 1810s); publications peaked at more than 4,500 in 1808 (with nearly 17,000 holdings) but would not return to similar numbers until 1815. During the revolutionary period and into the 1840s, total publications surged from around 9,000 books to 14,093 in the year 1848 (with a corresponding growth in libraries holding copies), immediately dropping nearly a third to 10,550 in the year after the tumultuous the 1848 revolutions. The WorldCat data even present a smaller but palpable falloff in book publishing activity much earlier, during the Thirty Years’ War: a publication rate of almost 2,000 books a year from 1617 to 1620 falls to only 974 in 1627, to an abysmal 517 in 1639, and no real recovery until well after the war's end in 1648.

German data also show publication rates that may correspond to positive historical effects on the national literary scene. During the reign of Bismarck, the rates of publication within German-speaking lands surge from 14,404 in 1871 (Bismarck's ascension to Imperial Chancellor) to 28,440 in 1890 (the year of his resignation). The German publication data even experience an early peak during the Protestant Reformation that began in 1517. Only 327 publications survive from 1517, but that leaps immediately to 777 in 1518 and crests at 2,341 in 1523. This peak also could reflect the historical importance of Reformation materials, including the explosion of warring theological pamphlets, which libraries would tend to heavily collect and preserve.²⁴

The unique nature of the German data also is revealed in comparison to the historical data from France (figure 3). The French data also contain a drop in publication rates in the revolutionary year of 1848, from 7,262 publications in 1847 to 5,815 in 1849. Even more pertinently, in 1871, the year the Prussians occupied Paris, the publication record crashes from 9,269 in the previous year to fewer than 6,000. The years 1789 and 1790, around the time of the French Revolution, see the most dramatic spike in both French national print publication and, perhaps tellingly, in libraries’ collection and preservation of materials, visible in figure 3 as a quadrupling of printed publications collected by libraries to more than 20,000 titles.

Alternatively, one can compare the twentieth-century data from Germany (figure 1) to the twentieth-century data from neighboring Poland (figure 4). The Polish data, as might be expected, do show a slump during their occupation by Nazi and Soviet forces, as well as a surprising dip in general worldwide holdings in the late 1970s, followed by a surge in publishing after the fall of Communism, a period that also corresponds to a very volatile period in the Polish publishing industry as the vibrant underground publishing community from the 1980s suddenly experienced the collapse of state censorship as well as turbulent economic conditions.²⁵ Also, one can note that the national holdings for Poland—holdings of Polish publications in Polish libraries—are in every year consistently lower than the total number of titles, an indicator of fewer data from libraries within Poland.

For a non-European country like South Africa, on the other hand, a comparable slump in publication activity during World War II would be less expected, and indeed it is less evident in the data. One major turning point in the South African data profile (see figure 5) instead follows the establishment of a national legal library deposit system in 1959.²⁶ South African publications, as reflected in the WorldCat data, exploded from 1,695 publications in 1959 to more than 9,500 in the year 1990. However, international pressure in the 1970s and 80s against Apartheid could be reflected in a tension between flatter growth in South African books within library collections outside of South Africa: over the same span of years, South African books discovered by this research in South African libraries increased more than tenfold, from 3,104 South African library holdings in 1959 to 41,234 in 1990, while non-South African library holdings of the same publications only increased from 7,000 to 20,000. The numbers are even more striking in proportion to the overall surge in publication rate. In 1959, each book published in South Africa was collected on average by at least four other libraries worldwide; by 1990, the proportion was fewer than two libraries per publication.

Reliable data on language of publication (see appendix A) also emerged from most of the sample national datasets and in most cases reflected appropriate and expected differences between them. Although the dominant language of publication in books from Bolivia was expectedly Spanish, note the presence in WorldCat of books published in indigenous languages, such as Aymara, Quechua, and Guarani. The data from Chile can be treated with even greater trust knowing that the country enjoys a national shared library network.²⁷ English-language publishing is apparently stronger in Chile than in Bolivia, despite the lesser dependence of the Chilean data on libraries external to Chile and perhaps can be explained by the dominant strain of minority English speakers in Chile.

For the larger European countries, conversely, the book publishing seems to be more Eurocentric, with less emphasis on a single dominant national language, greater emphasis on publishing in other European languages, and fewer appearances of minority but native languages such as Danish, Frisian, and Sorbian in the German publications, and Lithuanian in those from Poland. On the other hand, these data include a strong presence of books published in Latin in the data from Germany and Poland—a concrete reflection of Western libraries’ function as historical memory institutions. Not shown in appendix A but definitely present in the Polish and German data are works in Greek, Hebrew, and historical languages such as Low German and Middle High and Old High German, again reinforcing the function of the library collection in preserving these aspects of the historical culture. The data from South African book publishing, on the other hand, appropriately reflect the more complex linguistic heritage of that country—two dominant languages of the ex-colonial powers (and likely founders of much of the South African publishing industry) with a very healthy dose of both translations into, and works original to, a variety of indigenous languages.

The language data also can be parsed out by year across the historical span of a dataset. German-language publications in Germany, for instance (see figure 6), display interesting nodes around a spike in German-language publications in 1913 (leaping to more than 60,000 publications in the national language) as the country was gearing up for what would become the First World War, and a surprising dip through the 1970s. Compare that to the very different graphical shape of Germany's Latin-language publications (figure 7). Germany's Latin-language publishing first spiked in 1517 (with 245 Latin-language publications collected and preserved by libraries at the outset of the Reformation and appearing in WorldCat), and remained generally strong (around 1,000 titles per year published in Latin) up to the nineteenth century. At the start of the twentieth century, however, Latin-language publishing in Germany fell off to numbers around 800 titles per year; in the 1930s, during the Nazi era, this fell even more dramatically, and 1968 (the year the Second Vatican Council liberalized many Roman Catholic practices) ushered in another decline in Latin-language publication. (The spike in the year 1972 could be an outlier data point within the trajectory of the data; a random sample of Latin-language publications from Germany in 1972 revealed no explanation for such an increase in Latin-language publication in this year.)

Data mining also was able to reveal patterns in book translations (see appendix B). Such data, allowing a view of the interactions between languages used by a culture, is among the most valuable to UNESCO as it seeks to measure cultural statistics. Once again, the data for Poland and Germany reflect a more Eurocentric vision, with translations from a variety of major European languages into German and Polish responsible for the majority of the translations; although in these data as well, translations into English from German and Polish also figure prominently. Latin and Greek translations also are represented in the WorldCat data from these two countries. The data on these translations were, however, somewhat less prominent. Greek to German translations were the tenth most prevalent, at 6,178 books; in Poland, Latin to Polish translations (2,687 books) were somewhat less common than Polish to German, and Greek to Polish lagged behind, although still recording 1,072 translations. The data on translations for Bolivia, Chile, and Thailand give a similar picture of both the predominant languages spoken in each country (Spanish and Thai), as well as the interaction of English and other languages with these dominant languages.

The data for the more culturally diverse country of South Africa, on the other hand, presents a much more varied tapestry of translation data, including translations between English, German, Afrikaans, and a number of tribal languages. Even more interesting historical patterns are revealed by plotting these data on a similar historical graph (figure 8). Whereas book publications from South Africa (figure 5) were more avidly collected after the 1959 deposit library system, the dominant type of translated works during the early decades of the Republic (1961 until the mid-1980s) are translations between English and Afrikaans, varying between around 25 and 150 per year. After the Zulu and Xhosa languages were decriminalized in 1991, however, translations from them enjoy a surge to more than 50 translations per year.

A different set of shifting cultural interactions may be viewed in the historical patterns of translated works from Poland (figure 9). The most overt pattern is the large increase in translations from English, French, and German after the fall of Communism; English to Polish translations, particularly, rocketed from fewer than 400 in 1989 to more than 6,000, literally off the chart, after 2000. But from the 1950s to the 1980s, the most important pair of languages tended to be Russian and Polish. In a lesser-populated dataset, such as the publication and collection profile from Thailand (figure 10), the spiky nature of the graph suggests that data mining in this instance is nearing the functional end of its reliability. The Thai data on translations also are the only instance in which translations into English lead the statistics (as reflected in WorldCat).