At the beginning of the quarter, the staff at the Getty sent several spreadsheets including one of every record from the Sales Catalogs database that contained the word “frame.” Elizabeth Walmsly’s article, “Italian Renaissance Paintings Restored in Paris by Duveen Brothers, Inc., c. 1927-1929,” an early reading assignment that features notable restorers and frame makers, shows how important their roles were in selling art as the market expanded in the mid- to late 19th century. Art historians have generally not paid much attention to frames, but in Art and Market Studies, they become impossible to ignore. For this project, I wanted to study how frames were described and how that may have been related to activity in the market.
Authentic and Beautiful Frames
This project began with a word cloud based on the contents of the entire spreadsheet on frames. Certain words caught my eye, such as “curious” and “ancient,” “antique,” “old,” and “original.” Most of the other words in the word cloud were somehow related to the beauty of the frame, characteristics that would help sell a painting. According to Walmsly, the importance of frames to the art dealer Sir Joseph Duveen was that they helped make frames saleable, and he preferred frame makers such as Ferruccio Vannoni for their artistry.1 Paintings are elements of interior decoration, so the frame has to flatter both the painting itself as well as the room it is placed in. In a letter to Vannoni in 1929, Duveen expressed these considerations in regard to the repeated reframing of a painting that just would not sell: “The frame is too overpowering for this picture and for most rooms…[y]ou must therefore design a very beautiful frame.”2 So why would dealers note a frame’s age?
Rarity is one characteristic that makes a luxury good desirable. In the introduction to the exhibition catalog Italian Renaissance Frames, Timothy J. Newbery, et al., describe how paintings were routinely reframed as fashions changed and frames became damaged. Also, the new owner of a painting would often reframe it to match the other paintings in his gallery. Old frames, especially Italian Renaissance frames, consequently became very rare and desirable such that frame makers began producing pastiches and forgeries, or restored them to improve their condition.3 Adjectives that refer to a frame’s age and authenticity (“old,” “ancient,” “antique,” and “original”) appear in the sales catalogs of art dealers because these qualities made a painting saleable, just like the beauty of a frame.
Les cadres in the Stock Books of Goupil & Cie
During the field trip to the Getty Research Institute, the archivists talked about how the dealer stock books contained a lot of information about frames. In particular, the art dealership Goupil & Cie kept records on the price of framing, whether a painting was sold with or without a frame, if they sold a frame on its own, if a frame was repaired, and so on. I had originally intended to explore the Goupil stock books to study the restoration practices of old, authentic frames and its effect on the value and marketability of a painting. Unfortunately, searching for the keywords “cadre” and “encadre” (“frame” and “framed” in French), even with a wildcard asterisk after the “r,” only finds 81 records. Something was amiss, because I expected to find many more records. After searching the database further, I noticed that sometimes the Dealer’s Cost line on the website contains the phrase “cadre compris” with a price. These records did not show up in my earlier searches, and searching for the keywords “cadre compris” returns only one record, so it appears that the search engine does not index every column in a record. This line of research had to be put aside due to time constraints, so I turned my attention back to the Sales Catalog.
Working with the Sales Catalogs Database, and Histograms of Adjectives
The data from the Sales Catalog database lent itself to comparisons between frames that were desirable either for their beauty or their authenticity. I decided to make a histogram to provide a more detailed and precise visualization than a word cloud. The data describing frames is unstructured, so I had to quantify the use of words somehow in Excel. The first step was to concatenate all of the columns with the word “frame” in them so that their contents could be searched. Then I made columns for each word in the word cloud. I saw other adjectives in the merged column that did not exceed the word cloud software’s threshold, so I found as many extra adjectives as I could and added columns for them. Using a search formula, I was able to check each record for the presence of the relevant terms. Then the COUNTIF function added up the results. Once I worked out this technique, it was quite easy to prepare new datasets. This approach was not completely clean, however. Sometimes these adjectives described something other than the frame, such as “old” being used in reference to “an old head (portrait),” or “fine” in reference to the painting rather than the frame. With nearly 6000 records, it was impractical to clean the data, so I left it at that.
I made two histograms with the resulting data: one with no categories, and another with color-coded categories for beauty, authenticity, wood type, and shape. These are more comprehensive and precise than word clouds, but not more accurate.
Next, I made another histogram to see how many frames per year used words that speak to the authenticity of a frame. I wanted to answer the question: “When did authenticity become a desirable characteristic?” The dataset for this visualization was small enough to clean, so I removed all of the records where the adjectives old, ancient, antique, and original did not refer to the frame. I also removed four very early references from 1689-1690 that made a large, empty gap in the histogram. The histogram shows that dealers began to regularly note a frame’s age during the early 19th century. Curious as to whether there were any changes in the preferred word to describe such frames, I made another histogram with each word on its own. This was not clearly readable because the columns were so narrow, so I turned it into a line graph instead. While the dataset is rather small, the line graph suggests that “old” and “ancient” were more common than “antique” and “original.”
Italian Renaissance Frames
In order to confirm the desirability of Italian Renaissance frames, I made a pie chart of the nationalities of artists whose frames were “authentic.” The first chart shows that the frames of Italian and Dutch paintings were the most common, followed by British paintings. To compare the percentages of authentic frames with the larger market, I also made a second pie chart for paintings with “beautiful” frames. About 17% of the beautiful frames were Italian, while about 32% of the authentic frames were Italian. Old Italian frames were rare at this time, so their relatively common presence among authentic frames might be an indicator of high demand. Dutch frames are equally common among authentic frames, but they occupy roughly equal market share in beautiful frames as well. In other words, they were not disproportionately represented among authentic frames.
Distribution of Sale Prices
To get a sense of whether paintings with authentic frames were more valuable than beautiful frames, I made two scatter plots with the year of the sale on the x-axis and the price on the y-axis. One scatter plot only had authentic frames, and the other had both authentic and beautiful frames. Since the prices are not adjusted for inflation, these visualizations can only show whether the value of authentic frames increased or decreased faster than beautiful frames. I did not take into account other factors such as the size of the painting or the reputation of the artist, either. Looking at the scatter plots, it seems that authentic frames were not more or less valuable than any other frames. The scatter plots do show that most of the volume at auctions were low priced paintings, and that there were several price points. Looking at the first scatter plot, the prices are concentrated around two pounds and under, five to seven pounds, and 10 pounds and above. The higher price ranges are visible in the combined scatter plot, where there are layers for about 10-40 pounds, 60-120 pounds, and 150 all the way up to 340 pounds.
Specialists in Authentic Frames
Lastly, I wanted to see if anybody was particularly interested in selling and buying paintings with authentic frames. I made an alluvial diagram that links seller to buyers with stripes whose thickness is determined by how many paintings were exchanged between them. Many sellers and buyers are unknown, suggesting that either dealers often did not record who sold or bought a painting, or many buyers preferred to be anonymous. It does not seem like there were any particularly avid sellers or buyers of paintings with authentic frames. A few sellers sold more than one painting with an authentic frame: J.M. Lamb, Robert Berks, Rev. James Ireland, John Arden, J. Meredith, and Lord Saye and Sele. The Lord sold a few paintings to Redfern, which is the only relationship that stands out.
In the future, it would be interesting to return to the Goupil stock books and to work with the Duveen X-Books. I could export every year of records in the stock books and search for all of the ones that mention frames, especially the ones that contain “cadre compris” in the Dealer’s Cost line. A fuller dataset from the Goupil stock books would make it possible to figure out how much money dealers spent on framing paintings, and to see if dealers lost money on reframing as a way to keep stock “fresh” like they did with paintings circulated among dealers. It would also be possible to compare the average amount of time that it took to sell a framed vs. unframed painting. Lastly, the Goupil and Duveen stock books could reveal their frame restoration practices.
As for the Sales Catalogs database, one could go deeper into the questions raised here or ask new questions. There is the matter of why there were those four very early outliers from 1689-90 on the histogram of authentic adjectives. Is the gap caused by the database’s coverage, or was there a change in the significance of age? Another issue that came up is whether the demand for Dutch paintings was different from Italian paintings. Furthermore, the prices of ebony frames versus “Dutch black” frames could be compared to see if paintings with “imitation” materials cost less than those made with real ebony. The sales catalogs might also reveal changing fashions and tastes in frames, perhaps by correlating the use of broad frames to the number of paintings at the Salon, or by making line graphs of the different materials, techniques, and shapes, then tying that in with their sale prices. Maps could also show where certain styles and characteristics were popular. For example, there might not have been any individual sellers or buyers who specialized in authentic frames, but maybe there was an auction house or city that did. While cleaning data, I also noticed that some paintings went to auction multiple times. A dataset with records of such paintings, not limited to those that mention frames, could give insight on speculative investment in art, for instance. On a related note, this project has only considered frames from the perspective of a subset of the entire database. Putting the paintings with notable frames into the context of all of the paintings that do not mention frames allows for comparisons that are not otherwise possible.
Newbery, Timothy J., George Bisacca, and Laurence Kanter. Italian Renaissance Frames. New York: The Metropolitan Museum of Art, 1990.
Walmsly, Elizabeth. “Italian Renaissance Paintings Restored in Paris by Duveen Brothers, Inc., c. 1927-1929.” Facture 1 (2013): 58-77.