How Old Is My Painting?

A signature and a style can be imitated in an afternoon. A stretcher cannot. The reverse of a painting is where the trade and the conservation studio both begin, because everything back there was made by an industry rather than an artist: timber cut to a standard, canvas woven on a particular kind of loom, fixings bought by the box, and labels applied by dealers, framers, exhibitions and shippers who all had reasons to date their own work.

Photograph the back before you touch anything, corners included, and photograph the edges where the canvas turns over the stretcher. Those tacking edges are frequently the only part of the original surface that has never been cleaned, varnished or restored.

Canvas woven by hand has an irregular thread count and visible variation across the weave; machine weaving is monotonously even, and once you have seen the two side by side the difference is obvious. Linen dominates until the twentieth century, when cotton — cheaper, whiter, more elastic — becomes ordinary for artists' canvas. A cotton support on a picture presented as eighteenth-century is a serious problem for that story.

Panels carry their own evidence. Hand-sawn boards show irregular saw marks and are usually thicker; machine sawing is regular. Plywood belongs to the late nineteenth century onwards, and hardboard — the compressed wood-fibre board sold as Masonite — was invented in 1924, so a picture painted on it cannot predate that. Commercially prepared canvas boards and academy boards are similarly nineteenth-century-and-later products.

A fixed strainer — four members with no means of tightening — is the older and cheaper solution and persists on modest work of every period. An expandable stretcher, with slotted mitred corners and small triangular keys tapped in to tension the canvas, is a nineteenth-century refinement onwards. The presence of keys is not a date on its own, but a stretcher that has never had them, on a canvas that has plainly been retensioned, tells you the canvas and the stretcher are not the same age.

Then read the fixings. Hand-forged and cut nails give way to machine-made wire nails at the end of the nineteenth century. Staples are a twentieth-century convenience and, on the tacking edge, close to conclusive: a staple-stretched canvas is not an old canvas, though it may well be an old painting on a new stretcher.

The collapsible metal paint tube was patented in 1841 and changed what paintings look like: colours became consistent, painters worked outdoors, and impasto became easy to lay on thickly. Ground-in-the-studio paint behaves differently and often shows uneven pigment dispersion under magnification.

Individual pigments are the sharpest tool of all, because each has a date of introduction. Prussian blue appears at the start of the eighteenth century, synthetic ultramarine in the 1820s, titanium white in artists' paint in the 1920s, phthalocyanine blue in the 1930s. Any one of them in a picture claimed to be earlier settles the matter — and this is laboratory work, not eyesight, which is precisely why serious attributions include a materials analysis.

Craquelure — the network of fine cracks that forms as paint, ground and support age at different rates — develops over decades and follows the mechanics of the support beneath it. On canvas it tends to run in a rough grid; on panel it follows the grain. It is not damage so much as a clock, and a picture presented as three hundred years old with an unbroken, glassy surface is asking to be explained.

Deliberately induced cracking exists and looks wrong: too regular, too shallow, evenly distributed where a real network varies with the thickness of the paint. Under a raking light — a lamp held almost parallel to the surface — you can also read the texture of the ground, the ridges around filled losses, and whether the impasto has been flattened by an old lining.

The best evidence on many paintings is glued to the back of it. Colourmen stamped their canvases and their stretchers, and those firms have documented trading dates and addresses, which brackets the canvas to a period of years. Gallery, exhibition and shipping labels do the same, and old auction lot numbers can often be traced through printed catalogues.

Read them for anachronisms as well as for dates. A postcode or a zip code, a telephone number in a modern format, a barcode, adhesive tape rather than glue and paper: each pins a lower bound on when the label was applied. And treat every inscribed date on the front with caution — dates were routinely added later by family, dealers and optimists.

Before dating a painting, confirm it is one. Under a ten-power loupe, a printed reproduction resolves into a regular pattern of dots or an even ink film, while paint has brushwork with direction and thickness. Canvas transfers — a print laminated to canvas and varnished, sometimes with brushstrokes added in gel — are common in the market for “old” pictures and are obvious under magnification and raking light.

This matters for age because printing technologies date themselves cleanly. A half-tone dot pattern, an inkjet's fine spray, a lithographic grain: each belongs to a period, and none of them belongs to a seventeenth-century workshop.

Dendrochronology can date a wooden panel by its growth rings, often to a range of a few years, and it tells you when the tree was felled rather than when the picture was painted. Pigment analysis, X-radiography and infrared imaging answer the questions the naked eye cannot. All of it is available through conservators, and for anything that might be significant it is money well spent before an auction estimate rather than after.

A photograph, meanwhile, does something more modest and useful: it can propose a style, a school and a period from the image alone, which is how most people work out whether the picture on the wall deserves a conservator's afternoon at all. The Painting Identifier does that in a few seconds; the rest of this article is what it cannot see, because the evidence is on the other side of the canvas.