Nov 23 2011 11:15am

The Late Mike Fellinger’s Turkey Algorithm: An Exercise in Geek Cooking

I still think of this as Jon Singer’s Turkey Algorithm, because that’s the title under which I knew it for decades; but Singer says it was devised by the late Mike Fellinger, and must be credited to him. It goes:

“For a turkey of greater than ten pounds, the roasting time should be equal to 1.65 times the natural log of the weight of the bird in pounds, cooked at 325 F.”

If you’re not a person who normally calculates natural logs, go to Google. Say you have a 20-pound turkey. Type in natural log 20 and hit the search button. Google will tell you that the natural log of 20 is 2.99573227. Multiply this by 1.65. The result will be 4.9429582455, or five hours.

And because there is no geekery without pilpul:

Ken Houghton’s Condensed Version: Just FYI, you can put the whole thing into Google. Typing in ln 20 * 1.65 (for a 20-pound turkey) yielded ln(20) * 1.65 = 4.94295825.

Erik V. Olson’s Cavil: I have to dissent from Mr. Singer’s algorithm, because we can do this vastly better emprically. The correct way to cook a turkey involves a probe thermometer inserted into the thickest part of the breast. Cook until 161F, remove, cover, let rest for 15 minutes, uncover, carve. ... Technology. It’s a lovely thing.

Henry Troup’s Expanded Principle: My best-cut algorithm is a 3/2 power law (which is a direct inversion of the square-cube principle). By my cooking log, Jon’s algorithm gives a bit longer time than I use. However, I cook at 350F, so the coefficient shifts.... I log every (big) bird I cook to refine the technique. The numbers in cookbook are a bit on the fuzzy side for my tastes. The same general algorithm can be used for things like pig roasts, too, just shift the curve to match the species and done-ness needed.

Brooks Moses’ Special Circumstance: If you have a small free-range turkey, your ideal cooking time may be considerably shorter than that yielded by the algorithm. Use a thermometer. 

The Original Entry and Comment Thread: Making Light, 22 November 2007.

1. strongdreams
4 hours for a 12 pound turkey? Insane. Brine the bird over night, then roast at 450F for 30 minutes (to get a nice crisp skin), then cover the breast with foil (to prevent over-browning) and set the oven at 350F until the breast meat is 161F. Total time for a 12 pound bird is 1-3/4 to 2 hours. (Method from Alton Brown, the best food scientist I know.)
2. Yatima
Can I just say how thankful I am for discussions of algorithmic approaches to roasting? Truly you are my people.
Ben HM3
3. BenHM3
The article made me smile, then the grin broke out in laughter at strongreams (#1) wise quoting of the coolest dude in cookery, but Yatima's remark (#2) made my day.

I'm thankful for you all.
Teresa Nielsen Hayden
4. tnh
Exactly so, Yatima. You too, Ben. Happy Thanksgiving. May your joys be fractal and

Strongdreams, it's nice that Alton Brown (who is indeed cool) has published a fully debugged version of the Alton Brown method, but that doesn't mean it's the only way to do it. Are you going to claim there's neither geek value nor good eating in turduckens, or deep-fried turkeys? Furthermore, if there isn't a mad alchemist somewhere out there who's rigged a sous vide turkey and is cooking it even as we speak, I'll eat my lab coat.

If you want to get really picky about it: first, as Brooks Moses could tell you, the algorithm can get a bit shaky at its lowest ranges. Second, you're brining the turkey, you're starting it at 450 F., and you're thereafter roasting it at 350 F., none of which match the algorithm. You get results that don't match the curve? Funny thing, that.

Happy Thanksgiving to you as well.
5. JoeNotCharles
tnh @4: The point is not that this is a "better" algorithm for the same type of turkey cooking, as Erik Olson and Henry Troupe's versions are. The point is that this is a better way to cook the turkey!
Joris Meijer
6. jtmeijer
Teresa, there is always the bionic Turkey described at the 'Cooking Issues' blog. Which is cooked whole at low temperatures in oil, with the leg bones replaced with tubing to ensure proper cooking of the leg-meat.

Legs 1 hour at 66 C by circulating hot duckfat/butter. Followed by the whole bird in the fat for 1.5 hours at 65 C. Finished by crisping the skin.
7. strongdreams
I don't think there is much likelihood that 30 minutes at 450 and 90 minutes at 350 somehow equates to 240 minutes at 325. The algorithm seems excessive, at least for lower masses.

There is certainly a mathematical beauty to incorporating ln into cookery. Before I accepted it as trustworthy, I would want to see the curve tested at several points, and if it truly is shaky at low mass, then it's not really a logarithmic relationship, and hence only the illusion of beauty. But then all beauty is illusion anyway...

Subscribe to this thread

Receive notification by email when a new comment is added. You must be a registered user to subscribe to threads.
Post a comment