The painting was the wrong size.
Not by much — three centimetres in height, perhaps two in width — and not in any way that would register to a casual eye or a distracted one. But Mara Selles had been looking at paintings since she was seven years old in her grandmother's house in Tarragona, where a small seventeenth-century Dutch interior had hung in the kitchen and been used, variously, as a clock, a calendar, and a source of instruction on the correct preparation of fish, and she had been looking at them professionally for fifteen years, and she knew, in the particular way she knew things before she could explain them, that this canvas was not the dimensions that the documentation said it was.
She did not say this immediately. She set down the ultraviolet lamp, removed her magnifying loupe, and looked at the painting from five paces with her arms crossed and her expression giving nothing away. The painting was a Northern European interior, attributed to a follower of Pieter de Hooch, circa 1660: a woman at a window, a tiled floor, a letter on a table, light entering from the left with the particular quality of Dutch coastal morning that no forger had ever successfully reproduced because it was not a technique but an understanding. The attribution was not Old Master first tier — it was the kind of work that sold steadily at specialist auction for mid-range prices among informed collectors, its value less in the name than in the quality, and its quality was considerable.
She crossed back to the easel and looked at the lower left corner.
The palazzo studio was quiet. Outside, the canal moved with its perpetual soft insistence, and somewhere above her the building's ancient framework expressed a minor opinion about the afternoon. The light through the arched window was good — west-facing, the September sun still at a height that produced the raking angle she preferred for surface examination. She had been given the room to herself for the afternoon, a key on a velvet cord, and a folder of documentation that was comprehensive in the way of documents assembled by someone who had thought carefully about what to include.
She looked at the lower left corner with the loupe.
The paint surface in a two-centimetre radius around the corner showed micro-craquelure consistent with the rest of the canvas — age cracks running in the shallow, irregular pattern of genuine drying over time. But the craquelure in the extreme lower margin, at the very edge, had a different formation. Slightly more regular. The lines a fraction too even. The kind of evenness you got from artificial ageing performed with technical skill on a time constraint.
She put the loupe away and reached for the UV lamp.
Under ultraviolet, the canvas told a different version of what it was: the varnish fluoresced in the greenish-yellow film of age, uneven in the organic, convincing way, and the paint layers beneath showed the normal stratified complexity of century-old work — except in the lower margin, where the fluorescence broke slightly, a thin line of cooler response running parallel to the bottom edge, perhaps four centimetres from it. A repair, possibly. A lining seam, possibly. Or a strip of later canvas incorporated into the original stretcher in a way designed not to be seen.
She photographed it with three different filters and then straightened and looked at the painting from a normal viewing distance.
It was beautiful. That was the problem with very good forgeries — not that they looked wrong, but that they looked right in a way that should have required the full weight of the original's history. The woman at the window had the absolute stillness of a subject who had been understood rather than observed, and the light on the tiled floor had the depth of a man who had spent years learning how Vermeer and de Hooch thought about morning — not copying them but inheriting them, the way a musician inherits from composers they have never met. Whoever painted this had known the original.
Mara turned to the documentation folder and opened it again, this time to the provenance page.
The chain was clean on its surface: private Italian collection, 1948; sold at small Milanese auction, 1971; acquired by Emilio Cantor, 1988; bequeathed to the Brancati estate on Cantor's death, this year. Each transfer documented. Each document plausible. The 1948 entry — private Italian collection, acquired in the immediate postwar period — was the kind of phrasing that appeared frequently in European provenance chains of that era, meaning almost nothing while appearing to mean something, and she had learned over fifteen years to treat it as a variable to be resolved rather than a fact to be accepted.
She wrote in her notebook: Lower margin anomaly. Canvas dimension discrepancy, est. 3cm H / 2cm W vs. documentation. UV: possible later addition at lower edge. Provenance: 1948 entry unverified by primary source.
She was writing the fourth line of notes when the door opened.
