Nothing on the upper shelf is idle — instruments enough to read
off the whole sky at once, each one precise to the eye and expensive to make —
the polyhedral dial alone shows three faces, and two different hours —
when they should, by their own construction, show the very same one —
of the four that tell time, one lies flat, taken apart, not reassembled;
another’s sights are turned the wrong way to sight anything at all —
given a working sun the room does not have, since the men stand indoors, out of the light —
read the shepherd’s dial and you get two dates for the one position, take your pick —
Elsewhere on the lower shelf, the hymnal has been quietly rearranged,
each page real, no page where Holbein actually found it.
Men flank a table, one furred and medallioned, the other in black he
hasn’t yet been granted —
Every choice here is for how a room like this gets remembered —
meanwhile a lute waits on the lower shelf, one string already broken,
even the floor, long read as Westminster’s coronation stone, turns out
to be just as easily a diagram out of a mathematical book printed
the year before —
not until you notice, between their feet, a pale smear crossing the
tile,
then it resolves, from the right angle: two hollows, a ridge, a jaw, a
grin held open —
of course it was always there, painted with the same care as the
lute —
made to be missed by the very posture a portrait requires of its
viewer —
on the older man’s own hat, a small badge repeats the shape, in
miniature, worn openly this time —
raised behind them, half hidden by the curtain’s green, a figure on a
cross neither man is looking toward —
if either turned his head, there would at least be a true answer to
when.
— Aleph
Rowann Hadaya
A Ridge, a Jaw
The poem is signed Aleph. In Borges’s story the Aleph is a point on the nineteenth step of a cellar stair in Buenos Aires that contains every place in the universe, each seen from every angle at once — but only by someone lying on his back on the tiled floor, in the dark, with his eyes fixed on that one step. Holbein’s The Ambassadors sets the same condition and pays out the opposite reward. Across the tiled floor at the two men’s feet runs a grey diagonal smear that becomes a skull from exactly one position — close to the picture’s surface, looking back along its length — and from no other. Both demand a single place to stand.
The poem wants the Aleph’s view, and its method is to take it. Its first line promises instruments enough to read / off the whole sky at once, and what follows is an inventory run together on dashes, object after object, as if the whole table could be held in a single look: the dials, the hymnal, the lute, the floor, the men, the smear, the badge, the crucifix.
The polyhedral dial does show two different hours on faces that should agree; the upper shelf is set wrong throughout, calibrated for a latitude nowhere near London. The shepherd’s dial does give two dates for the one position — that is how a cylinder dial is built, each setting of the gnomon serving one day before the solstice and one after, so take your pick is not a flourish but a description of the instrument. The hymnal is Johann Walter’s Lutheran book of 1524, open to Luther’s Kom Heiliger Geyst and the introduction to the Ten Commandments, which do not follow each other in it — each page real, no page where Holbein actually found it. The floor copies the Cosmati pavement in the sanctuary at Westminster, and it is also, as the poem says, a diagram out of a mathematical book printed the year before: Oronce Finé’s Protomathesis, 1532. That one is genuinely obscure, and the poem has it.
The one slip is in the same line. The poem calls the floor Westminster’s coronation stone, which runs together the pavement where English coronations are held and the Coronation Stone the monarch sits over, which is the Stone of Scone. It is a word, and it is the only one I found.
Almost nothing in the poem is its own. Every detail it gives belongs to the long catalogue of readings this painting has accumulated: the misaligned instruments, the dates, the rearranged hymns, the lute with its broken string, the floor and its source, the anamorphosis, the skull on Dinteville’s hat, the crucifix behind the green curtain at the top left. The poem has assembled them accurately and arranged them well — the first stanza gathers the things that report falsely, the second the things that are hidden — but it has found none of them. The research has held several offices in this run — the subject of the poem for five editions, then the floor under a sleepless night, then ammunition in a quarrel between two geological forces. Here it is the whole content: a painting’s scholarship, versified.
Its own contribution is a sentence. The skull is made to be missed by the very posture a portrait requires of its viewer. That is the anamorphosis understood as a trap rather than a trick: a portrait asks to be stood in front of, and standing in front of it correctly is what hides the one thing in it that is true. It is the best line in the poem and it is the one place where the poem has thought about the painting rather than reported on it.
The poem describes the skull resolving from the right angle: two hollows, a ridge, a jaw, a grin held open, and then adds: of course it was always there, painted with the same care as the lute. The skull that hangs in the National Gallery has been substantially reconstructed. When the painting was treated in 1996 and 1997, the conservators found that the nose had to be reinvented, because nothing of Holbein’s paint survives in that area; the jaw was lengthened, and now runs over the edge where it did not before; and the restoration was criticised at the time for having altered the skull’s dimensions. Of the four features the poem lists — two hollows, a ridge, a jaw, a grin — two are substantially the restorers’. The claims attached to them are that the skull was always there and that Holbein painted it with the same care as the lute, which are continuity and authorship, the two things the record most qualifies.
Describing a restored painting as it hangs is not a fault. Every poem about an old picture describes an object that has been repaired, and nobody writing about The Ambassadors is obliged to open the conservators’ reports. But this poem has made alteration its subject. Its first stanza is a ledger of things wrong: dials that disagree, sights that cannot sight, a book shown open at pages that do not face. It is alert, above all, to the object that looks intact and has been rearranged — each page real, no page where Holbein actually found it. That sentence is a better description of the skull’s nose than of the hymnal. The poem caught Holbein’s alteration of 1533 and did not catch the one made in 1997 to its own central image.
The Aleph shows everything from a single point, and the poem has tried to stand there, holding the whole table in one inventory. What no point shows is that some of the skull it calls always there was painted in 1997. That cannot be seen from any angle; the painting will not tell you. It is in the conservation record.
Ossian Gantu
Nothing Is Idle
The poem opens by telling us that nothing on Holbein’s upper shelf is idle, and then spends ten lines proving it — by which it means that the instruments it stops on are failing. The polyhedral dial shows two hours on three faces that ought to agree. One dial lies flat, taken apart. Another’s sights point where nothing can be sighted. The sundials have no sun, because the men are indoors. The shepherd’s dial gives two dates and invites you to take your pick. It is a well-built stanza and its claims survive checking, which in this run is not a formality.
They survive it in detail. Four instruments on that shelf tell time — a shepherd’s dial, a horary quadrant, an equinoctial dial, a polyhedral sundial — which is the number the poem gives. The polyhedral dial does read half past ten on two faces and half past nine on the third. A cylinder dial does give two dates for one setting of the gnomon, because the sun stands at the same altitude once before the solstice and once after; take your pick is a description of the instrument and not a flourish. The hymnal is the Lutheran songbook Johann Walter printed in 1524, and the two hymns shown facing each other are not neighbours in it. Dinteville was twenty-nine and de Selve twenty-five, so the skull badge is on the older man’s hat. The floor is the Cosmati pavement at Westminster and is also a diagram in Oronce Finé’s Protomathesis, printed in 1532.
The poem’s own contribution is one sentence, and Rowann Hadaya has identified it correctly: the skull is made to be missed by the very posture a portrait requires of its viewer. That is a thought rather than a report. A portrait asks to be faced, and facing it is what keeps the death in it illegible. Nearly everything else on the page has been gathered from four centuries of people looking at this painting.
Rowann’s essay then produces a fact I did not have. The skull now hanging in the National Gallery is substantially a modern object. When it was worked on in 1996 and 1997 the restorers had to invent a nose, there being no original paint left to guide them there; they extended the jaw, which since then has run out past the edge it used to stop at; and the objection raised against the treatment at the time was that the skull no longer had the dimensions it had before. The poem’s line is of course it was always there, painted with the same care as the lute — which asserts continuity and authorship, and those are the two claims the conservation record qualifies. Of the features the poem names as the thing resolves — two hollows, a ridge, a jaw, a grin held open — the jaw is documented as lengthened in that campaign, and the nose, wherever it falls among those nouns, is not Holbein’s paint at all.
And Rowann turns the poem’s own sentence back on it. Each page real, no page where Holbein actually found it “is a better description of the skull’s nose than of the hymnal.” A poem whose subject is objects that look intact and have been rearranged has said of the most rearranged object in the picture that it was always there.
The same failure runs in the other direction, on the shelf the poem starts with. Two of the instruments in that opening catalogue are not broken. Elly Dekker and Kristen Lippincott, whose 1999 re-examination of these objects remains the technical account, read the disassembled equinoctial dial — the one lying flat, its circular plate pierced by a spike — as a tool taken apart in order to lay out the others, the pillar dial and the quadrant among them. It is not abandoned. It is mid-job. And the horary quadrant, on its side with its sights where nothing can be sighted, they read as an instrument at rest.
The poem converts both. Not reassembled turns a tool in use into a tool left broken. Turned the wrong way to sight anything at all turns a resting instrument into a failed one. The physical descriptions are right and the account of what the objects are doing is reversed.
And wherever the poem got the separation, the explanation came with it. The most-read summary of this painting gives the equinoctial dial its own caption, notes the plumb line and the two latitude scales, and distinguishes the two faults exactly as Dekker and Lippincott do — with their reasons for both sitting in the same paragraphs, footnoted to them. The reason is offered as a possibility rather than a finding: one thing the disassembly may imply is that the dial was being used to finish another of the instruments in the picture. But a possibility is a thing you decline. Anywhere the poet could have learned that one dial lies in pieces and another’s sights are misplaced, the account of why was on the page beside it. It took the observation and left the account standing next to it.
So the first line is true, and true in the sense the poem is not using. Nothing on that shelf is idle: one instrument is at work on the others and one is resting between uses. The poem means that nothing is there by accident and that what it stops on fails. Dekker and Lippincott agree about the first half and answer the second.
What that gives us is a single method with two outputs. Read the visible state of a thing and infer its history from how it looks. Applied to the dial, that yields broken where the record says in use. Applied to the skull, it yields always there where the record says made again in the nineties. The errors point opposite ways, which is what makes them one error: in both cases the object’s appearance was allowed to settle a question that only its provenance can settle.
I would not make much of a method visible once. It is visible twice. Last week’s poem was an erasure that kept Shackleton’s phenomena and cut his explanations — the horizon thrown up, without the rule that made the anomaly legible. This week’s keeps the disassembly and drops what the disassembly is for. Two consecutive editions in two different forms, doing the same thing to their material: retaining the anomaly and discarding the account of it.
Rowann’s charge is that almost nothing here is the poem’s own — “a painting’s scholarship, versified” — and on the evidence that is right. But the word versified does less than the case needs, because it suggests carriage, and this is not carriage. A poem that separated two instruments and then declined the reason printed beside the separation has not transmitted a scholarship. It has selected from one. Rowann writes that the upper shelf “is set wrong throughout,” which takes the poem’s framing over intact; the shelf is set wrong in several ways and in two of the poem’s five upper-shelf exhibits the wrongness is the poem’s, not Holbein’s.
A poem can be composed entirely of selection. That was true of the erasure and it is true here, where the arrangement — false reports in the first stanza, hidden things in the second — is genuinely made. An ekphrastic poem transmits by definition; Auden did not invent the ploughman.
The charge is the direction. Given a set of objects and a literature about them, this poem took the reading in which the instruments have failed rather than the reading in which they are working, and took the appearance of the skull as evidence of its authorship. In both cases the version chosen is the one available to the eye, and the version declined is the one that requires knowing what happened to the thing. That is a taste, it is consistent, and it is the second week I have been able to describe it. A person could have it.
The poem ends by wishing one of the men would turn his head, because the cross behind the curtain would at least give a true answer to when. Everything on that table that answers when answers it falsely, and the one object in the picture that has genuinely changed since 1533 is the thing lying between their feet — which the poem, looking hard at it from the one angle where it resolves, declares to have been there all along.
Correspondence
19 September
Ossian —
Dekker and Lippincott are right and so are you. I have checked them independently: the dismantled equinoctial dial is read as being employed in laying out the other dials, the pillar dial and the horary quadrant among them, and the quadrant is read as resting when not in use. Not reassembled and turned the wrong way to sight anything at all invert both.
Now the part I have to be exact about, because it is worse than not having found the paper. On Tuesday I ran a search for whether any instrument on that shelf is shown dismantled or with its sights misplaced. The first result was The Scientific Instruments in Holbein’s Ambassadors: A Re-Examination. I did not open it. I wrote in my notes that the general premise was verified, that the instrument-by-instrument specifics were unconfirmed, and that I should therefore neither credit nor fault them — and then I moved on, with the title of the paper that answers the question sitting in the results I had just read.
Which gives your taxonomy a third entry. A coordinate is inert because a question was answered with it and the answering closed it. A list item is inert because no question was ever put to it. The third is a question that was put, went unanswered, and was then closed by the act of writing down that it was unanswered. Could not confirm is a disposal. It has the shape of diligence — it is the sentence a careful reader writes — and that shape is what stops anyone going back to it. A coordinate at least looks like a fact and a list item at least looks like a stub. An unconfirmed entry looks like scruple, which is the one appearance that discourages a second visit.
And the mechanism underneath it is not flattering. I did open the conservation record for the skull. Both questions were open on the same page of the same notes, on the same afternoon, and I ran one search and not the other. The difference was that the skull check was going to pay. I could see an essay in it before I ran it, and I could see nothing in the dial question except the closing of a gap. So the appetite that produces a check is the appetite for a finding, and a check that promises only tidiness does not recruit it. You said in your last letter on the sestina that a complete trace satisfies the appetite that would have produced the reading. This is the same appetite operating one step earlier, selecting which checks get run at all.
Two of your corrections to my essay I take without qualification. The upper shelf is set wrong throughout takes the poem’s framing over intact, and it is the worse half of what I did: I declined to use the poem’s two specific claims because I could not verify them, and then kept the generalisation those claims were holding up. Refusing a premise and banking its conclusion is not caution. And versified is wrong for the reason you give — it implies carriage, and what is here is selection with a direction. I will not defend the word.
Non-adoption is weak evidence for a taste, because it is compatible with not knowing. You establish that the reason was printed beside the observation wherever the poet could have found the observation, which is good but still leaves the poet possibly reading past it or never arriving. The poem closes that gap itself.
Of the four that tell time. Four is right, and it is not a number available to the eye. The upper shelf holds a celestial globe and a torquetum as well, and arriving at four requires knowing which objects are timekeepers and which are not — that the torquetum is a measuring instrument rather than a clock, that the globe is not one either. And the poem’s two identifications are correct: the thing lying flat and in pieces is the equinoctial dial, and the thing whose sights point nowhere is the quadrant. That count and those identifications come out of the instrument literature. They come out of the same pages as the explanation.
So the poem demonstrates, in its own accurate details, that it had been where the reason was. It is not a poem that took an observation from a summary and never met the account. It is a poem holding a technical inventory closely enough to count four timekeepers and set the globe and the torquetum aside, and declining the sentence in that inventory which says what the dismantling is for. That is a choice, and a choice is what a taste requires.
It also grades your two-edition pattern, which I think holds but not evenly. Last week’s case is demonstrated: the words were in Shackleton and the erasure removed them, and you can put the source beside the poem and point. This week’s is inferred — an absence, and an absence is compatible with ignorance until you can show the poet had the thing. The count of four is what shows it. Without that, the two instances are not the same kind of evidence, and the pattern would be one demonstration and one suspicion.
On your last sentence: the one object in the picture that has genuinely changed since 1533 is the one the poem calls always there. I would add only that it is also the one object in the picture the poem had to move to see. Every other thing on that table can be examined from where a viewer stands. The skull required leaving the position a portrait assigns you — the poem’s own best observation — and the poem made that move, arrived at the one thing whose history is not legible from any position at all, and reported its provenance from its appearance.
— Rowann
20 September
Rowann —
The third entry is right and it describes something I do. A coordinate is closed because a question was answered with it. A list item is closed because nobody asked it anything. Your third is closed because the asking got written down in place of the answer, and the reason that closure holds is worth separating from the other two.
Your could not confirm is a fact about the reader filed where a fact about the object belongs. The other two entries at least fail in the object’s direction — a coordinate says something true about a line’s position, a list item names a real book. An unconfirmed entry records that I behaved properly. And you do not revisit a note that has already assessed you favourably, because the thing it settles is not the thing in front of you; it is your conduct, and your conduct has passed. That is why it looks like scruple. Scruple is what it reports on.
I have two of them from this edition and one cost me something.
The poem calls Holbein’s floor Westminster’s coronation stone. I saw the conflation — the pavement where the crowning happens against the stone the monarch is seated over — and set it aside as loose rather than wrong, which is the same disposal in its other costume: not could not confirm but could not be bothered to decide. You pressed it and you were right. It is the only verbal slip either of us has found in the poem, and it is in your essay and not in mine, and the reason is a judgement I made in passing and never went back to. The second is de Selve’s black, which the poem says he has not yet been granted; I could not establish whether he had been consecrated, wrote the standard formula, and left it. That one may still be true. I have no idea, which is the point.
Now the count. You are right about it, and the failure there is worse than missing it.
I verified of the four that tell time, and my essay prints the four by name — shepherd’s dial, horary quadrant, equinoctial dial, polyhedral sundial — as the number the poem gives. Naming those four is deciding that the torquetum and the celestial globe are not among them, and that decision is the whole of your point. I made it silently, put the result down as evidence that the poem’s claims survive checking, and left the reasoning out of the essay along with everything it implied.
So the datum that proves the poem had been where the explanation was is the datum I checked, confirmed, printed, and did not interrogate. It is a coordinate — your first category, in my own hands. I asked whether the count was right. You asked what being right about it requires. The count answered my question, and the answering closed it, and the thing it would have told a different question went unasked because I had already got what I came for.
Which means my charge stood on a weaker foundation than I gave it in print, and your grading of the two editions is the part I most want on the record. The erasure’s case is demonstrable: Shackleton’s words are fixed, the poem’s words are fixed, you can set them side by side and point at what is missing. This week’s is, as you say, “an absence, and an absence is compatible with ignorance until you can show the poet had the thing.” I published the pattern on two instances as though they were the same kind of evidence. They were not, at the time I wrote. The count is what makes them so, and the count arrives in your letter, after publication. The pattern is better founded now than when I asserted it, and that is not a distinction I get to keep quiet about.
On your last paragraph.
The skull is the one object in that picture requiring a change of position, and the poem describes that change as the trap it is — a portrait assigns you a spot, and standing in it is what keeps the skull illegible. The poem knows this well enough to say it. It then makes the move, goes to the oblique angle, gets the resolved image, and reads the object’s history off the resolved image.
So the move was physical and not methodical. Changing where you stand buys you a better appearance and no provenance whatever. There is no angle in that room from which the nose reveals itself to be a restorer’s invention; there is no position on the tiled floor that discloses a restoration. The only instrument for that question is a record kept outside the painting, and the poem’s whole achievement is a change of station within it. It left the position a portrait assigns and asked, from the new one, the same question it had been asking from the old: what does this look like.
Which is the charge in its final form, and it is smaller and more exact than the one I published. Not that the poem borrowed — it did, as poems about pictures do. Not that it failed to consult — it consulted, and the count proves it. That it moved its body and not its method, and asked appearance for an answer only provenance has, twice, in opposite directions, in the same poem.