Rot of Ages


On Nov. 28, I wrapped eight lumps of raw barley dough in fig leaves and put them in loosely lidded plastic containers. Forty days later, I mixed the now-rotted barley with flour, salt and water and set the result in a warm spot to rot for two months more. Then I ate some.

I was just following the medieval Arab recipes for a sauce called murri. Really. Rot some barley, then rot it some more; that’s what they tell you to do. You could also mix the rotted barley with milk instead of water and make a sort of relish called ka^makh.

I’d made my first batch of rotted barley about 10 years earlier. At the time, I asked a medical researcher whether there might be anything dangerous in these rotted barley condiments, and he warned me that they were probably infected with the common mold Aspergillus flavus, which produces dangerous carcinogens called aflatoxins. So after tasting the result of that first experiment, I washed my mouth out about 70 times.

The sauce I got at the time was pleasant but not exactly exciting. It didn’t seem so attractive that I could imagine cooks going to all the trouble to make it, much less overlooking the rather ghastly process by which it’s made. I wasn’t at all sure I’d got it right.

So last year I decided to try another batch of murri. This time, I sent the results of my rotting to a laboratory to find out whether there was indeed anything dangerous in it (see “OK, It’s Rotted . . .,” below).


Regular Food section readers have already met my eight lumps of barley dough, which I was rotting in three batches according to the slightly contradictory recipes I’d found in medieval manuscripts. We did the first stage of rotting in the Times Test Kitchen--very far from the food preparation area, you may be sure.

The first of the three batches consisted of five lumps of dough crowded in a single box. In these rather moist conditions, the barley rotted vigorously. After three days, the lumps had greenish spots and were covered with a cloud of translucent threads, indicating that molds of the genus aspergillus had taken up living quarters. We nicknamed these lumps Spot, Whiskers, Einstein, Skinhead and Johnny Rotten, and after 40 days they were mottled all over in black, white, charcoal and about four shades of green.

The second batch was rotted in drier circumstances, only two lumps to the box. Kate Moss and Captain Picard, as we called them, developed a greenish tinge and a few tiny white patches but no cloud of threads.

Finally, one lump was rotted under ashes, as a 10th century recipe suggests. This guy, Pig Pen, also rotted only slightly.

After 40 days, I ground the lumps up and made each batch into a paste with flour, water and salt. The recipes say to rot the paste on your rooftop in the middle of summer, so I put the three batches on a heating pad on the low setting, which kept the mixtures at about 95 degrees. I stirred and moistened them all daily for 40 days. Finally, I added more water and waited another 10 days before extracting a liquid sauce.

Here’s what we got.

Kate Moss, Captain Picard and Pig Pen turned pinkish tan (Pig Pen with an ash-gray tinge). The unappealing “liquid” sauce I got from those batches was like salty glue with a sickly-sweet aroma. By contrast, the totally rotten batch (Spot, Whiskers et al.), which started out green, quickly turned a deep mahogany brown. At first it smelled like rotting leaves or a damp basement, but within 10 days it smelled distinctly like . . .

Soy sauce.

Yes, soy sauce. The sauce extracted from it had a comparatively earthy (some said muddy) texture and flavor, but there was no doubt about it: Murri was a medieval Mediterranean soy sauce.

Big surprise; I had always assumed that soy sauce had to be made from beans. But the usual Japanese soy sauce recipe is 50% wheat, and it turns out that oats, millet, rye and even barley are sometimes added in China and Korea. In fact, there’s no reason you have to use any beans at all. All you need is proteins and carbohydrates, which are also found in grain.

The process is basically the same whether you’re making soy sauce or murri. In the first stage, the moist dough or soybeans become infected with molds, which can grow where conditions are too dry for yeasts or bacteria. The molds break down starches to sugars and proteins to amino acids.

In the next stage, water and salt are added. Now yeasts and bacteria take over--not just any yeasts and bacteria, of course, only those that can live in a fairly salty environment (soy sauce is 18% salt; the murri paste is 25% salt by weight before water is added).

At this point, a lot of chemical reactions start. Some carbohydrates are turned into acids, including glutamic acid, which is related to MSG. The yeasts produce alcohol; it combines with the various acids to make the aromatic compounds called esters. The result is the sweet, appetizing soy sauce flavor. The longer it ages, the stronger the flavor.

There are a couple of differences. Soy sauce is more flavorful than murri because it’s made from beans, which contain more proteins than grain, and it’s aged longer: in Japan, a minimum of 18 months. And in East Asia, a “starter” of rotted rice is added to the soybeans, rather than just setting out what you plan to rot and letting any old molds move in. In Japan, the making of this starter (koji) is a highly technical activity done in laboratory conditions.

On the other hand, many households in East Asia still make their own soy sauce using wild molds. Often people wrap the rice for the starter with leaves of certain plants, which are likely to have mold spores on them: reeds in China, rice straw in Korea, banana leaves in Vietnam, hibiscus leaves in Indonesia. Fig leaves evidently served the same function in the Arab recipes.

Soy sauce and murri apparently evolved from brine sauces. Everybody knows about the East Asian fish sauces known as nuoc mam or nam pla, and it’s also common knowledge that the Greeks and Romans loved a fish sauce called garum. All these sauces were merely the brine from pickled fish.

In ancient China, brine sauce (jiang) was made from meat as well as from fish (imagine using the pickling juices of ham or corned beef as a sauce). A “starter” (qu) of moldy rice was regularly added to the brine to give more flavor.

By about 2,100 years ago, the Chinese were making a soybean sauce. Or paste, rather, because they didn’t start pressing the rotted soybeans to extract a free-flowing liquid sauce until about 450 years ago. The idea of pickling beans might not have occurred to a Westerner, but the Chinese were already pickling fruits and vegetables in brine. Anyway, they had a long tradition of culturing soybeans with mold; Chinese black beans start out as yellow soybeans.

The medieval Arab sauce must have arisen from the Mediterranean fish sauce. The word al-murri (which the Arab dictionaries say was often pronounced with one “r”: al-muri) may come from the Greek word halmuris, which just meant “a salty thing.” Salmuria, the Latin word for pickling brine, comes from the same Greek word. How rotted barley entered the picture is a little mysterious, but it seems to be a Persian contribution. The Arab word for rotted barley (bu^dhaj) comes from Persian, as does the name of the paste (bunn) from which murri is extracted.

If the Mediterranean and East Asian sauces were unrelated, they were certainly used the same way, to provide a salty, aromatic flavoring to cooked dishes. In Spain and North Africa, every banquet had to include a dish called “the dish of murri” (laun al-murri), which seems the equivalent of the Chinese stews known as red-cooked meat.

So the mystery of what murri was seems to be solved. But it raises another mystery: Why did murri die out? Apparently it hasn’t been made for 400 or 500 years. Some people had always found its color dark and unpleasant, but that shouldn’t have mattered much if the flavor was liked. Maybe tastes just changed--the ancient Greek fish sauce was also dying out at the same time.

And maybe the story might have been different if Mediterranean cooks had started making those richer sauces that result from beans. We might be eating soy-seasoned falafel today.


This recipe is from a 13th century Spanish Arab cookbook known as the Manuscrito Anonimo. It remarks that chickpeas could be added and that the vinegar is optional.

1 onion

2 tablespoons cilantro

1 pound boneless beef chuck, cut in 1 1/2-inch cubes

1/4 teaspoon salt

2 teaspoons ground coriander



1/2 teaspoon ground ginger root

1/4 teaspoon ground cloves

1/4 cup dark soy sauce

1 1/2 tablespoons vinegar

1 1/2 tablespoons oil

1 1/2 cups water

Chop or puree onion with cilantro. Put meat in pot and add onion mixture, salt, coriander, 1 teaspoon cinnamon, 1/2 teaspoon pepper, ginger, cloves, soy sauce, vinegar, oil and water. Bring to boil, cover and simmer until meat is tender, about 1 1/2 hours. Sprinkle with cinnamon and pepper to taste.

2 to 3 servings. Each of 3 servings:

256 calories; 1,784 mg sodium; 68 mg cholesterol; 13 grams fat; 7 grams carbohydrates; 28 grams protein; 0.40 gram fiber.


The recipe is from “The Key to Chinese Cooking” by Irene Kuo (Knopf, 1977). The turnips are particularly delicious.

3 tablespoons oil

2 cloves garlic, peeled and lightly crushed

1 pound boneless beef round or chuck, cut in 1 1/2-inch cubes

2 tablespoons rice wine or dry Sherry

4 tablespoons dark soy sauce

2 teaspoons sugar

1 whole star anise

1 green onion, cut in 4 pieces

2 quarter-sized pieces peeled ginger

2 cups boiling water

1 pound turnips, peeled and cut in eighths

1/2 teaspoon salt

Heat heavy stewing pot or casserole over high heat. Add oil and swirl 30 seconds. Add garlic and stir, pressing against pan with spoon. Add meat and brown on all sides quickly. Add wine, soy sauce, sugar, star anise, green onion and ginger, then stir. Add boiling water. When mixture returns to boil, reduce heat and simmer, covered, 1 hour. Add turnips and salt and simmer until meat is tender and turnips plump and moist, 30 to 40 minutes.

3 to 4 servings. Each of 4 servings:

258 calories; 1,425 mg sodium; 51 mg cholesterol; 15 grams fat; 10 grams carbohydrates; 21 grams protein; 0.94 gram fiber.