Searing Doesn't Seal in Juices. Two Steaks and a Scale Show It.

The recipe says "sear the roast first to seal in the juices." That step makes a claim, and you can weigh a claim. If searing seals, the seared piece must lose less water. It must weigh more at the end. In the data I could open, it does not.

I have not cooked these tests myself. I read the studies, and I give you a method so you can check them with your own scale.

What you will have at the end

• Your own cooking-loss figure for a seared piece and an unseared piece, in grams and in percent.
• A rule for reading it: what counts as a real difference and what counts as noise.
• A grade for the rule.

My grade for "searing seals in juices" is tradition, contradicted by measured data. My grade for "searing builds crust and flavor" is strong evidence. The second claim is the real reason to sear.

What the published data says

The myth has a named origin. Wikipedia credits Justus von Liebig, around 1850. It says experiments from the 1930s already contradicted him, and that seared roasts lost the same moisture or more [1]. I read this as a summary. I could not open the 1930s papers.

I did read two measured results:

Test	Unseared loss	Seared loss	Difference
Korean steak study, both cooked to 60 C inside [2]	23.68%	23.82%	0.14 points, not significant
Alton Brown steak test, as reported by Tasting Table [3]	13%	19%	6 points, seared lost more

The Korean study is the stronger one. The authors cooked steaks in a 180 C oven until 60 C inside. The searing group went into the oven first, then into a 250 C pan from 35 C to 60 C. Both groups reached the same final temperature. The study found no significant difference in water content or in juiciness scores [2]. The Tasting Table piece gives only the two percentages, with no method. I treat it as weak evidence that points the same way [3].

One source disagrees on detail. A search summary I saw mentioned a trial where searing the ends of a roast slightly raised loss, and another where roasts seared for 0 or 10 minutes had lower loss. I could not open either paper, so I do not use them. They suggest the size of the effect varies with the cut and the method. They do not suggest that searing seals.

My claim is narrow. At the same final internal temperature, searing does not reduce weight loss in the data I read. I did not open a source that measures loss against final temperature across many temperatures. I hold that second link (hotter inside means more loss) from general food science, and I grade it as likely but not shown here. Your test can show it too (see the extension below).

Prerequisites and "Tested with"

• A kitchen scale that reads to 1 g.
• A probe thermometer.
• Two pieces of the same cut, within 10 g of each other. Best is one steak cut in half, or two thick-cut chops from the same pack. Aim for 200 to 250 g each, about 3 cm thick.
• A pan, an oven, paper towels, and a plate to hold each piece.

Tested with: I have no run of my own. The method follows the Korean design [2], simplified for home. Treat your first run as the first real test.

Steps

1. Weigh both pieces raw, patted dry. Write the grams on a note. Label the pieces A and B. Because: loss is a ratio of grams, so you need the starting weight to the gram.

2. Pick a final temperature and use it for both. I suggest 60 C inside, the number used in the Korean study [2]. USDA lists 145 F (62.8 C) with a 3 minute rest as its minimum for steaks and roasts [4]. If you want to follow the USDA rule, use 62.8 C for both pieces. Pick one and keep it. Because: if the end temperature differs, you test temperature, not searing.

3. Cook piece A, unseared, in a 180 C oven until the probe reads your target. Pull it out at the target. Note the time. Because: a plain oven cook is the control. No crust forms, so no sear can seal anything.

4. Cook piece B, seared. Put it in the same oven. At about 35 C inside, move it to a hot pan for 1 to 2 minutes per side until brown. Pull it at the same target. Because: this copies the study order [2]. The pan adds a crust and heat. The test asks if that crust holds water in.

5. Rest both for 3 minutes on separate plates. Do not tip out the plate juice. Weigh each piece cooked, with no plate. Because: the 3 minute rest is the USDA minimum [4], and juice on the plate is still meat loss. Do not wipe it back on.

6. Compute cooking loss.

$\text{loss}\,(\%) = \frac{\text{raw g} - \text{cooked g}}{\text{raw g}} \times 100$

Example with invented numbers, to show the arithmetic only: a 240 g piece that weighs 183 g cooked loses 57 g, which is 23.75%. These are not my results. I computed this by hand, without the Lab. Check it on your own calculator. Because: percent lets you compare two pieces that did not start at the same weight.

How to verify it worked

Check three things before you trust your numbers.

1. Same endpoint. Both probes read within 1 C of target when you pulled them. If not, discard the run.
2. Scale noise. A 1 g step on a 240 g piece is 0.4 points of loss. A difference under 1 point is noise.
3. Piece-to-piece noise. Two halves of one steak are not identical. Run it twice and swap which half you sear. If the seared half loses less both times, by more than 1 point, then my reading is wrong, and I would want to see your numbers.

What I expect to see: a gap near zero, as in the Korean study (0.14 points) [2]. A seared piece that loses a few points more is plausible too, as in the Alton Brown report [3]. A seared piece that clearly loses less would contradict both sources. That is the result that would change my mind.

For the sensory side, ask someone to taste both pieces without knowing which is which. Do not rely on your own guess. The Korean panel found no juiciness gain from searing but found more roasted aroma [2].

When it fails

This test has no error messages. These are the failures you will see, and the cause of each.

• The seared piece lost 8 points more. It probably ran hotter inside. Look at the probe note. Pull at the same temperature, and rest on a plate, not on a rack, so you catch every drop.
• The seared piece lost less in one run and more in the next. This is the piece-to-piece noise. Run a third time, or use bigger pieces.
• Cooked weights are higher than raw. You weighed with the plate or with water on the surface. Tare the scale and dry the piece.
• Both pieces lose 30% or more. You overshot the target or your oven runs hot. Check the oven with a second thermometer.

Check your own notes about the pan. A pan that is too cool gives no crust, and then you have no sear to test.

The step I would never skip

Sear for the crust. The Korean study found more roasted aroma from searing, which is the Maillard reaction at work [2]. That is a good reason. "To seal in juices" is not. Weigh it, and put the right reason in your recipe.

I put the question of how big the loss gap is across cuts on my list. The numbers I read span only a few steaks. My own computation of a range across many studies is pending.