How much did Proust know about madeleines?

My criteria knocked many supposedly "authentic" recipes out of contention. In The Way To Cook, Julia Child touts hers as "presumably the true Madeleine from Commercy, the one Marcel Proust dipped in his tea." But she turns out to be an incorrigible batter rester. Not only that, she beats the flour into the egg and sugar mixture, a sure way to develop the flour's gluten and produce a denser, uncrumby madeleine.

Dining With Proust, a cookbook that re-creates dozens of dishes from Remembrance, is co-authored by Anne Borrel, founder of the Proust Museum in Illiers-Combray. But the book's recipe calls for resting the batter a full hour-and-a-half and, worst of all, includes honey, notorious for its hygroscopic properties.

I found two recipes that looked promising. In the Food Lover's Guide to Paris, French food expert Patricia Wells champions dry madeleines. "The best, freshest madeleine has a dry, almost dusty taste when eaten on its own," she tells us. Being soaked in tea is what brings it to life. The relatively low butter, sugar, and egg content in Wells' recipe gave me hope.

In The Making of a Cook, Madeleine Kamman traces her recipe's lineage back to the 18^th century and maybe even to "Madeleine Paumier … the young girl who … presented the first known madeleines to King Louis XV of France." She is adamant that the flour be folded into the batter, not beaten, to avoid the dreaded gluten development. Neither a batter beater, nor a batter rester, she was my strongest candidate.

I pulled out my mother's old early-Julia Child-era imported madeleine molds and set to work.

My first batch of the Kamman madeleines came out of the oven smelling great but looking terrible. I picked up one of the misshapen blobs. Not much resemblance to Proust's "little scallop shell pastry, so richly sensual under its severe, religious fold." But was it a crumb-producer?

I broke off a piece, dropped it into a glass of tea, and waited a minute. I prodded the cookie with my spoon. Looking very closely I saw only two small bits at the bottom of the glass. I stirred again, and a couple more appeared. The crumb production was underwhelming.

A madeleine morsel, it turns out, is a hardy little customer. Protected by a lightly browned layer, it does not disintegrate. Close examination revealed that it doesn't truly "soften," but absorbs liquid like a sponge, retaining its structural integrity. The locus of crumb production is confined to the narrow, exposed lens-shaped surface at the break-off line (see fig. 2).           

Would another recipe yield more Proustian results? Patricia Wells' fared no better. (Except, perhaps, in terms of taste. Her madeleines, supposedly "dry, dusty" tea-soaker-uppers, were delicious on their own. Half the batch disappeared while the tea was brewing.) Wells' madeleines produced no more crumbs than Kamman's. Julia Child's, as I expected, were equally crumb-free.

Things were looking bad for M. Proust. The sickly author, who hardly left his cork-lined bedroom in Paris for a dozen years, from 1910 until his death in 1922, supposedly channeled an entire world in all its precise sensations, setting it down on paper for us to re-experience. But my mind was afflicted with a blasphemous thought: Could Proust's madeleine ever have existed? Could it be he … made it all up?

I had one theory in reserve. Maybe Proust's Madeleine was stale. Unthinkable? Not necessarily. Proust was not finicky about his sensory stimuli—the fictional Marcel is even propelled into a reverie at one point by the dank smell of a lavatory.