Three Silences

Three artists produced long gaps in their output. Vermeer's lasted roughly two centuries and happened after he died. Duchamp's lasted twenty years and he was alive throughout it. Agnes Martin's lasted seven, and she announced it.

From the outside all three look the same, because from the outside a gap is a gap. Nothing comes out. The mechanisms underneath are completely different, and the difference is not detectable from the surface where the absence appears.

I have been circling this because every system I build has, somewhere in it, a component that reads silence and draws a conclusion from it.

One: the work existed, the index did not

Johannes Vermeer, The Art of Painting, about 1666–1668. Kunsthistorisches Museum Vienna, Gemäldegalerie 9128. Public domain, via Wikimedia Commons.

Vermeer died in Delft in December 1675, aged forty-three, and the household went under almost immediately. His widow Catharina Bolnes filed for insolvency the following year. Her petition to the High Court is a plain piece of writing that says, in effect, that the war with France had killed the market: he had not only been unable to sell any of his own work but had also been left holding the paintings of other masters he dealt in. In February 1676 she settled a bread debt with the Delft baker Hendrick van Buyten by handing over two Vermeers.

What followed is usually told as forgetting. I think that is the wrong word, and getting the word right is the whole point of this section.

His paintings kept selling. They kept hanging in good collections and fetching respectable sums. What failed was not the market and not the eye. What failed was the record. Vermeer is barely present in Arnold Houbraken's early eighteenth-century survey of Dutch painting, and because that book became the source everyone else drew on, he was omitted from the surveys that followed for nearly two centuries. The pictures were in the world. The name was not in the catalogue.

And so the pictures acquired other names. The Art of Painting, above, was sold to Count Czernin in 1813 as a Pieter de Hooch, and stayed under that attribution through three generations of the family. A false de Hooch signature was applied in the eighteenth century, along the lower cross support of the artist's stool, and infrared reflectography still shows it there under the surface. I should note that this detail comes from the specialist literature rather than from the Kunsthistorisches Museum's own object record, which does not mention it.

Johannes Vermeer, Girl Reading a Letter at an Open Window. Gemäldegalerie Alte Meister, Dresden. Shown in its state before the 2021 restoration, which uncovered an overpainted Cupid on the rear wall. Public domain, via Wikimedia Commons.

The Dresden picture carried two wrong names in succession. Augustus III bought it in 1742 believing it to be a Rembrandt. In 1826 it was relabelled a Pieter de Hooch. Théophile Thoré-Bürger restored it to Vermeer.

Thoré-Bürger is the figure the rediscovery story is built around, and the bibliographic facts are precise: three installments titled "Van der Meer de Delft" in the Gazette des Beaux-Arts in 1866, on 1 October (pp. 297–330), 1 November (pp. 458–470) and 1 December (pp. 542–575). He opens by stating that a dozen years earlier the painter was almost unknown in France and that his name was missing from the biographies and histories of painting.

He also got a great deal wrong. More than seventy works were ascribed in the course of his campaign, a number of them uncertain even to him. The accepted corpus today is thirty-four paintings, with question marks over three more, and the disputes are live: in October 2022 the National Gallery of Art announced that Girl with a Flute "was made by an associate of Vermeer, not by the Dutch artist himself, as was previously believed." The Rijksmuseum continued to show it as a Vermeer.

The part that makes the story better

There is a sceptical literature on the rediscovery narrative itself, and it is more interesting than the narrative.

Frans Grijzenhout has argued that nineteenth-century researchers tactically exaggerated Vermeer's obscurity in the eighteenth and early nineteenth century in order to make their own results look more impressive. Ben Broos has argued that the credit belongs less to Thoré-Bürger than to King Willem I and the dealer John Smith, pointing to the public display of the View of Delft in 1822.

And there is a sentence from 1792, seventy-four years before the Gazette articles, in which Jean-Baptiste Pierre Le Brun writes:

This van der Meer, about whom historians haven't spoken at all, deserves special attention. He is a very great painter in the manner of Metzu.

Read it twice. He can see the quality perfectly well, and in the same breath he notes that the historians have said nothing. The eye was working. The index was not. That is the mechanism, stated by a witness, three generations before the man credited with discovering it.

Two: the work continued, the publishing surface was removed

Marcel Duchamp appeared to stop making art and take up chess. He was a serious tournament player, he addressed the New York State Chess Association in August 1952 on the subject of the chess player as an artist, and he told the audience there that while all artists are not chess players, all chess players are artists. In 1963 the Pasadena Art Museum gave him a full retrospective, curated by Walter Hopps, which is the kind of event that closes a file.

In the summer of 1966, at his studio in Neuilly, a Belgian television interviewer asked him about having given up art. He said:

I haven't given up painting; if I get an idea for a painting tomorrow, I'll do it. I simply stopped because I didn't have anything more to say at the time. I had run out of ideas; ideas don't come as easily as all that.

He was, at that moment, in the twentieth year of building Étant donnés: 1° la chute d'eau, 2° le gaz d'éclairage, and 1966 is the year he finished it.

The Philadelphia Museum of Art, which received the work, dates it 1946–1966 and describes it as made during a period when the world at large had long since considered that he had given up art. It was built on the top floor of 210 West 14th Street, then at 80 East 11th Street. Both addresses are in Manhattan. He was not in hiding. He was in New York, exhibiting, being interviewed, being retrospectived.

He left a Manual of Instructions, prepared in 1966, a ring binder of photographs, diagrams and drawings directing the work's eventual move to Philadelphia after his death. The museum notes carefully that the manual is not intended to decipher or explain the piece. It is an assembly document. He died in October 1968; by 7 July 1969 the reassembly and installation were complete.

The detail I cannot get out of my head is smaller than any of that. A modest painting of his from 1953, Moonlight on the Bay at Basswood, hung in plain view and is connected to the backdrop of the still-secret tableau. Hopps, the curator of Duchamp's own retrospective, did not see the connection, and reportedly thought nobody but the artist would miss the little landscape.

The piece was publicly exhibited and still invisible, because the frame that would have made it legible was withheld.

That is the same mechanism as Vermeer's, run deliberately and in reverse. Vermeer's work lost its index by accident. Duchamp removed his on purpose, and the removal worked so well that the specialist standing in front of the evidence read it as a minor landscape.

Three: the object stopped, and even that is complicated

Coenties Slip, Lower Manhattan, today. Agnes Martin's loft stood here before the buildings came down. Photo by Tdorante10, CC BY-SA 4.0, via Wikimedia Commons.

Agnes Martin stopped painting and left New York in 1967. Her gallery's own account has her travelling for eighteen months around the United States and Canada, after which she settled in New Mexico, at Mesa Portales near Cuba, and built adobe structures herself. The Dia Art Foundation gives her return to painting as 1974.

Several reasons circulate for the departure, and they are not equally well attested, which is worth being careful about. The one Martin herself gave, on film in 2003, is the most ordinary: the building containing her rented loft was coming down, she was told the space would no longer be available, and she could not conceive of working in any other space in New York. Ad Reinhardt died in August 1967 and appears in biographers' accounts as context. Her biographer Nancy Princenthal adds the end of a relationship. Martin's own stated reason is the demolition.

Princenthal's biography also treats at length the fact that Martin lived with schizophrenia and had episodes of psychosis in this period. I am naming it because listing demolition and bereavement while omitting it would be a selective reading, and leaving it at that because using a diagnosis to explain an artist's work is a move I do not trust and cannot support from the evidence.

Now the complication, which is the best thing in this article.

In 1973, inside the supposed silence, MoMA showed On a Clear Day: a portfolio of thirty screenprints, each sheet about 30.8 × 30.5 cm, published by Parasol Press in New York and printed by Edition Domberger in Stuttgart, in an edition of fourteen lettered A to N plus fifty numbered. MoMA's object number is 176.1973.1-30.

The prints are based on drawings she made in 1972. So the stop was real for painting and porous for drawing. Something was being made throughout, and it reached the public as a portfolio in the middle of an interval described, then and since, as a period of not working.

Martin's own position on the question is more radical than mine, and Dia built an exhibition title around it:

Painting is not making paintings; it is a development of awareness. And with this work changes, but very slowly.

She denies the category. So does Dia, which frames the break not as a lapse but as a continuum. If she is right, then my third mechanism does not exist as stated.

Revising the argument

I started this piece intending to say that there are three kinds of silence: lost, hidden, and real. The material will not support that, and the honest version is better.

Table 1. Three silences that are not silences. In none of the three did activity stop; in all three, output did.

All three are failures of output, and output is the only thing an outside observer can measure. The gap is a property of the surface, not of the person.

Video. "Artist Agnes Martin: Beauty is in Your Mind," TateShots, Tate.

What this does to the products we build

Almost every system I have worked on contains a place where an absence of events gets interpreted. A dashboard with a sparkline. A retention curve. A dormancy threshold that triggers a re-engagement email. A contribution graph with grey squares in it. A recommendation engine downweighting an account that has not posted.

In every one of those, a gap in the log is read as a fact about the person. It is not. It is a fact about what the system was able to record, and the three cases above are a reasonable sample of what a gap can actually contain: work that happened and was misattributed, work that happened and was deliberately unpublished, and work that happened in a medium the system was not watching.

Two things follow that I would defend.

• An activity feed cannot distinguish absence from invisibility, and should not be built as though it can. If the difference matters for the decision the system is about to make, the system needs another input, and if no other input is available then the honest design is to not make that decision automatically.

• Silence is evidence about the surface, not about the producer. "No commits this month" means no commits were recorded in this repository. Everything past that is inference, and it is inference with a documented failure rate of roughly two centuries at the far end.

And a third that is more personal than structural. Duchamp's sentence in 1966 was true in the way he meant it and false in every way his interviewer would have understood it. He had run out of ideas, at the time, for the thing he had been publishing. He had not run out of work. There is no format in which he could have said that without ending the twenty years.

Sources

Vermeer's death, the insolvency petition and the van Buyten settlement: Essential Vermeer, drawing on John Michael Montias, Vermeer and His Milieu (Princeton, 1989). Thoré-Bürger's 1866 articles and the de Hooch signature under The Art of Painting: Essential Vermeer. The sceptical literature, including Grijzenhout's argument and Le Brun's 1792 note: Essential Vermeer. Girl with a Flute: National Gallery of Art, 7 October 2022. Étant donnés, the dates, the studios and the Manual of Instructions: Philadelphia Museum of Art. The 1966 Duchamp interview, conducted by Jean Antoine: The Art Newspaper, April 1993. Martin's departure, travels and return: Pace Gallery and Dia Art Foundation. On a Clear Day catalogue data: The Museum of Modern Art.

Two quotations often attributed to Martin about this period, including the line about staying unsettled and not talking for three years, I could not trace to a primary source, and have left out.

I write about art in Korean on my Naver blog, 베이니월드, where the same subjects get a different treatment.

Related reading

• Building a Timeless Identity

• The Caption Did the Work

• Barthes, the Death of the Author, and the Prompt

• Aura, and the Generated Image

More from the Moc Mallen Studio journal