Collecting at American Art Institutions, 1890-1908

    Raphael Sasayama & Megan Driscoll


    What factors affected the early development of American institutional art collections? How much influence did one private dealer have on these collections? And will common art historical narratives be born out by data on turn of the 20th century accession practices?




    The Development of American Public Collections

    We first approached the public collections and provenance databases with an interest in exploring the relationship between museum collecting practices and art historical narratives about the relationship between European and American art. Specifically, we wanted to know if the paintings museums were purchasing in the late 19th and early 20th centuries would be predominantly works by European artists, supporting the widely held belief that American art was inferior to European art before World War II and the postwar rise of American modernism and Abstract Expressionism.

    This notion was summed up neatly by Clement Greenberg in 1955. A key American critic who is often credited with bringing the Abstract Expressionists to prominence, Greenberg famously observed that, as of the 1940s, “This country had not yet made a single contribution to the mainstream of painting or sculpture.”1 And even as the canon has expanded and questions of quality have become less absolute, the perception that American art was at least far less influential than European art before the 1940s has lived on. For example, in a seminal 1983 book examining the political implications of the cultural rise of the U.S., Serge Guilbaut points out that painting simply wasn’t a weapon in America’s “cultural arsenal” before the 1940s.2 And in an essay accompanying the Whitney Museum’s 1999 exhibition The American Century: Art & Culture, 1900-2000, Lisa Philips reminds us that, prior to WWII, “While Americans customarily looked to Paris for the latest developments in art, Europeans did not look at American art at all.”3 Given how entrenched this perspective is in art history and criticism, we wondered if it would be borne out by the purchases museums were making before this moment.

    When we discussed this issue with Christian Huemer of the GRI, he pointed out that the Getty’s stock book data might also help us answer another question about the development of American museum collections: Did Knoedler, an influential late 19th and early 20th century dealer whose stock books are owned by the Getty, have an outsized influence on American museums by providing a significant portion of the works they were purchasing? Knoedler & Company is a New York-based dealership that was founded in the 1850s by Michel Knoedler, who had originally come to New York from France as a representative of the French dealership Goupil. By the 1860s, Knoedler had bought out Goupil’s interest and established his own business.4 In the catalog for Knoedler’s 150th anniversary exhibition, they claim to have played a “pivotal role” in the formation of American museums because they were the primary conduit for art to early private collections in the U.S., which in turn germinated many public institutions through donation.5 By looking at the Knoedler stock books and comparing them with other public collections information available through the GRI, we hoped to explore this claim.


    Making Sense of the GPI

    In order to focus the project, Huemer recommended we narrow our focus to around the turn of the 20th century. We thus initially selected 1890 to 1914, ending at the start of WWI, but after learning that the later Knoedler stock books weren’t yet added to the GRI’s online database, we narrowed our time frame to 1890 to 1908. This left us with two sets of data. First, we looked at all of the American institutions that had records in the public collections database from 1890 to 1908, honing in on the nationalities of the artists for each painting in order to address our first question. Second, we looked at the Knoedler stock books to see which institutions they had sold paintings to between 1890 and 1908, then compared that with the total number of paintings each of those institutions had accessioned during those years according to their records in the public collections database.

    Several difficulties arose during our attempts to obtain these datasets. To find records for which paintings an arts institution acquired from Knoedler, we first searched the Provenance of Paintings and Public Collections databases, assuming that they were created from the same source data. This approach led to confusing results, such as finding 198 records in the Provenance of Paintings database with “Knoedler” as the owner and “Frick” as the institution, while there are only 171 records in the Public Collections database with “Frick” as the institution. One would expect to find fewer results in the Provenance of Paintings database, or at most an equal number, since not every artwork has provenance information. Ruth N. Cuadra, the Business Applications Administrator at the GRI, explained the causes of the discrepancy. Public Collections was made in 1985 and last updated in 1987, while Provenance of Paintings was made in 1985 and last updated in 2001. They actually come from different sources. She illustrated with a concordance that there is some overlap between the two databases, but there may be a record for a certain painting in one database but not the other. Rather than creating a concordance to combine the records in both databases, we chose to make datasets for our institutions from the Public Collections database, because it was overall more complete.

    There were only a couple of small issues with preparing the datasets from Public Collections. The exported spreadsheets do contain information on the artist’s nationality and the year that a painting was accessioned, but that data is mixed in with the artist’s name and the accession number, respectively. Simple Excel formulas extracted that data into their own columns so that we could make our visualizations. For paintings that were made by multiple artists, we only used the nationality of the first artist named. A few of the accession numbers did not contain any date information, so we deleted those records. Some paintings were later deaccessioned, but we kept those records because we were only interested in their acquisition.

    To find out how many paintings that Knoedler sold to an institution, we ran searches on the Dealer Stock Books by entering the institution’s name as the buyer. This posed no problem for institutions that existed during 1890-1908, but it raised a question for the Frick Collection, since it was not established until Henry Clay Frick’s death in 1919. According to the Public Collections database, 42 of the paintings in the Frick Collection were acquired between 1890-1908. However, searching Knoedler’s stock books with “Henry Clay Frick” in the buyer field shows that he bought 111 paintings in that time period. To resolve the disparity between Mr. Frick and the Frick Collection, we searched for “Frick Collection” in the keyword field, which returns 31 records. While this figure accords more with the Public Collection figure, it is another example of the vagaries of collecting and cleaning datasets.

    We decided to show the number of paintings that institutions acquired from Knoedler versus other sources with a variety of types of visualizations. Comparing two quantities for a short time span lent itself to simple histograms and pie charts. We made three kinds of histograms: standard, stacked, and stacked percentage. The first two show the number of paintings that an institution accessioned, allowing for comparisons of their total volume of collecting in those years. Due to scale and the small number of paintings acquired from Knoedler by most of the institutions, it can be hard to see some of the data, so labels are necessary. The stacked percentage histogram shows the relative prominence of Knoedler for each institution and the smaller quantities more clearly. Pie charts are also appropriate for visualizing quantities as percentages of a whole, though they suffer from the same drawbacks as histograms when it comes to showing small percentages. One pie chart shows the percentage of paintings acquired from Knoedler versus other sources by all institutions, and another shows the breakdown of paintings acquired from Knoedler by each institution.

    Lastly, in order to visualize the data on artists’ nationalities to answer our first question about pre-WWII museum collecting practices, we first tried network graphs. With the nationality of a painting’s artist as the source, and the institution as the target, the networks graphs were either too dense or too sparse to find significant patterns. An alluvial diagram, however, more meaningfully visualized this data.


    Interpreting Our Results

    We made several observations about our data after completing the visualizations. Looking at the alluvial diagram of artist nationalities, we confirmed that most institutions were, as expected, collecting European painters. The largest groups were from, in descending order, France, Italy, the United Kingdom, and the Netherlands. But we did find it interesting that the Detroit Institute of Arts (DIA) was unique in collecting a significant enough number of American artists for the U.S. to appear above Spain on the diagram. Further explorations of the history of the DIA might reveal an early preference for American art. It’s also notable that the Cincinnati Art Museum was collecting some American painting.

    Note: Click any image to expand.

    This alluvial diagram shows the origins (defined as the nationality of the artist) of paintings collected between 1890 and 1908 by American arts institutions with records in the GRI provenance database. The dominance of European artists reinforces expected patterns of preference before WWII, but it is notable that the DIA was collecting a significant number of American artists even at this early date.

    Our stock books research showed that Knoedler made up nine percent of all of the paintings accessioned by our institutions between 1890 and 1908. This is a fairly significant proportion, given that they’re only one dealer. However, most of this came from two institutions: the Huntington, who purchased 70 percent of their paintings from Knoedler during this time, and the Frick, who purchased 73.8 percent. Both the Huntington and the Frick are collections built from wealthy private individuals and/or families who later turned their collections into arts institutions. This suggests that Knoedler’s influence at this time may have been founded on social relationships between Michael Knoedler and the Fricks and the Huntingtons. This is significant considering the large influence that both institutions and their collections have had over time.

    What percentage of paintings were acquired from Knoedler vs. other sources by all of the institutions in aggregate?

    What percentage does each institution represent out of the total number of paintings purchased from Knoedler?
    How many paintings did each institution acquire from Knoedler vs. all other sources?
    What percentage of the paintings that each institution acquired during these years were from Knoedler vs. all other sources?

    These pie charts reflect the aggregate data above, allowing viewers to focus on each institution individually:

    What percentage of the total number of paintings that Carnegie accessioned during these years was purchased from Knoedler?

    What percentage of the total number of paintings that the Art Institute of Chicago accessioned during these years was purchased from Knoedler?
    What percentage of the total number of paintings that the Detroit Institute of the Arts accessioned during these years was purchased from Knoedler?
    What percentage of the total number of paintings that Frick accessioned during these years was purchased from Knoedler?
    What percentage of the total number of paintings that Huntington accessioned during these years was purchased from Knoedler?
    What percentage of the total number of paintings that the Metropolitan Museum of Art accessioned during these years was purchased from Knoedler?
    What percentage of the total number of paintings that the Museum of Fine Arts Boston accessioned during these years was purchased from Knoedler?


    Future Directions?

    Our preliminary findings suggest many opportunities for further research. To continue to pursue the question of artist nationalities, it would be helpful to acquire more complete accession data directly from the institutions themselves, and extend the time period up through the 1940s or 1950s. This would both allow researchers to confirm that the pattern holds up with a larger dataset, and see if the hierarchies between individual countries change or stay constant. Also, researchers may consider expanding the list of institutions to major European institutions to compare European and American collecting practices. In particular, it might be fruitful to examine Parisian vs. New York collecting practices in the postwar period to delve deeper into the art historical narrative of changing patterns of influence, which not only relate to the rise of American painters but also to New York replacing Paris as the center of artistic production.

    Turning to the Knoedler data, expanding to original datasets from the museum may also allow researchers to break down all of the museums’ sources to see if another gallery was more influential, or to see differences in patterns between museums. We’d also like to gain access to the full set of data from the Knoedler stock books in order to expand our time frame, and gain access to the Duveen stock books that the GRI has but has not yet digitized. These expanded datasets might also allow us to combine our questions to see if there was a relationship between the preference for European artists and the influence of a European art dealer working in the U.S. Was Knoedler–or any other dealer, such as Duveen–such a force as a dealer that they significantly prolonged institutions’ preference for European over American art?

    Finally, it would be interesting to delve further into the reason behind the Frick and the Huntington’s tendency to make such a large percentage of their purchases from Knoedler. Is there a more “zoomed in” narrative of their relationships to be told?


    1 Clement Greenberg, “‘American-Type’ Painting (1955),” in Art and Culture: Critical Essays, Beacon Paperback 212 (Boston: Beacon Press, 1961), 228.

    2 Serge Guilbaut, How New York Stole the Idea of Modern Art: Abstract Expressionism, Freedom, and the Cold War (Chicago: University of Chicago Press, 1983), 174.

    3 Lisa Phillips, “America Takes Command: 1950-1960,” in The American Century: Art & Culture, 1950-2000, by Whitney Museum of American Art (New York: Whitney Museum of American Art in association with W.W. Norton, 1999), 12.

    4 Charles Henschel, “A Personal History of Knoedler,” in The Rise of the Art World in America: Knoedler at 150, ed. M. Knoedler & Co, Sam Hunter, and Melissa de Medeiros (New York: Knoedler & Co, 1996), 9.

    5 Sam Hunter, “Foreword,” in The Rise of the Art World in America: Knoedler at 150, ed. M. Knoedler & Co, Sam Hunter, and Melissa de Medeiros (New York: Knoedler & Co, 1996), 7.

    back to top