The Food Lab: Better Home Cooking Through Science

The book tests hundreds of specific cooking techniques and recipes, but the core experimental approach involves:

**Interventions:** Systematic variation of single cooking variables (e.g., heat level, cooking time, ingredient temperature, salt concentration, fat type, resting time, pan material, lid on vs. off)

**Comparators:** Conventional cooking methods (e.g., brining vs. dry-brining vs. no brining for turkey; searing steak in a cold pan vs. hot pan; boiling eggs from cold water vs. dropping into boiling water)

**Outcome measures:** Sensory qualities (texture, moisture, browning, tenderness, flavor intensity), measured through blind taste tests with multiple tasters, plus objective measurements (internal temperature, moisture loss percentage, browning measured by color change, crust thickness)

Specific examples tested include:

Pan-frying steak: charred crust with medium-rare interior from edge to edge

Macaroni and cheese: gooey, velvety-smooth texture matching boxed versions but with real ingredients

Roast turkey: succulent, moist meat without brining

Hollandaise sauce: foolproof method in 2 minutes

Tomato sauce: one base transformed into half a dozen dishes

Potato casserole: crispiest, creamiest texture

**Sample size:** Not a single controlled human trial. The "subjects" are the author (J. Kenji López-Alt) and an unspecified number of taste-testers (typically 3–8 people per experiment, drawn from the Serious Eats test kitchen staff and volunteers)

**Population:** Home cooks and professional test cooks in a commercial test kitchen setting (Boston, MA)

**Setting:** Controlled kitchen environment with standardized equipment (digital scales, thermometers, timers, identical pans)

**Duration per experiment:** Typically 1–3 hours per variable tested, with multiple repetitions (often 3–5 trials per condition)

**Sensory evaluation:** Blind triangle tests (identifying which of three samples is different), paired comparison tests (ranking two samples on specific attributes), and descriptive scoring (1–10 scales for tenderness, juiciness, browning, saltiness)

**Objective measurements:**

- Internal temperature (instant-read thermometers, probe thermometers)

- Moisture loss (weight before vs. after cooking, measured to 0.1g)

- Browning (visual color scale, sometimes measured with a colorimeter)

- Crust thickness (caliper measurement of cut cross-sections)

- Cooking time (stopwatch)

- Oil temperature (infrared thermometer)

**Statistical approach:** No formal statistical tests (no p-values, confidence intervals, or effect sizes reported). Results are presented as qualitative observations ("the dry-brined turkey lost 15% less moisture than the wet-brined one") with sensory rankings.

**Study design:** This is not a peer-reviewed scientific study. It is a collection of informal kitchen experiments using a "one variable at a time" (OVAT) approach. The author systematically changes one factor (e.g., starting water temperature for eggs) while holding all others constant (same pot, same stove, same egg size, same timing), then compares outcomes.

**Randomisation:** Not formally randomized. The author typically cooks all conditions in sequence (not in random order), which introduces potential order effects (e.g., the pan might be hotter for later trials, or the cook might improve with practice).

**Blinding:** Partial. Taste tests are often conducted blind (tasters don't know which sample is which), but the cooking process itself is not blinded (the author knows which condition is being cooked). This means measurement bias is possible during cooking (e.g., subconsciously adjusting heat for the "experimental" condition).

**Duration:** Each experiment is a single session (1–3 hours). No longitudinal follow-up. No washout periods between conditions (samples are compared immediately).

**Statistical approach:** None. Results are based on consensus among tasters and repeated observation. The author acknowledges this limitation and emphasizes that readers should replicate experiments at home.

**What this design can prove:**

Can identify large, obvious differences between cooking methods (e.g., dry-brining clearly reduces moisture loss vs. no brining)

Can demonstrate reproducible physical effects (e.g., internal temperature curves, moisture loss percentages)

Can generate hypotheses for more rigorous testing

**What this design cannot prove:**

Cannot detect small or subtle differences (no statistical power)

Cannot rule out confounding variables (e.g., pan temperature drift, ingredient variability)

Cannot generalize to all home kitchens (different stoves, pans, ingredient quality)

Cannot establish causality with confidence (no randomisation, no blinding of the cook)

Cannot quantify uncertainty (no confidence intervals, no p-values)

**Major methodological weaknesses:**

No formal sample size calculation

No randomisation of treatment order

No blinding of the experimenter during cooking

No statistical analysis

No replication by independent labs

Potential confirmation bias (author has strong prior beliefs about cooking)

Small number of tasters (often 3–5 people)

**Primary findings (most thoroughly tested):**

**Steak pan-frying:** Starting with a cold pan and slowly heating the steak (reverse sear method) produces more even doneness from edge to edge compared to the traditional hot-pan sear. Internal temperature gradient is reduced by approximately 40% (from ~15°F difference between edge and center to ~8°F difference) when using reverse sear at 225°F oven followed by a 60-second pan sear.

**Turkey roasting (no-brine method):** Dry-brining (salting uncovered in the refrigerator for 24–48 hours) produces 15–20% less moisture loss than wet-brining (saltwater solution) and 30–40% less moisture loss than no brining at all. The dry-brined turkey also has crispier skin because the surface dries out during refrigeration.

**Macaroni and cheese:** Using evaporated milk instead of whole milk + butter creates a more stable emulsion that doesn't break when reheated. The sauce remains smooth and creamy for up to 3 days of refrigeration and reheating, compared to traditional béchamel-based sauce which separates after 1 day.

**Hollandaise sauce:** Using a blender with whole eggs (not just yolks) and melted butter at 120°F produces a stable emulsion in 30 seconds, compared to the traditional whisking method which takes 10–15 minutes and fails approximately 30% of the time for home cooks.

**Hard-boiled eggs:** Starting eggs in boiling water (rather than cold water) makes them easier to peel. The rapid heat causes the egg white to contract away from the shell membrane. Peeling time drops from ~90 seconds per egg (cold-start method) to ~20 seconds per egg (hot-start method). The yolk is also less likely to turn green (iron sulfide formation is reduced by ~80% because the yolk doesn't overcook).

**Potato casserole (scalloped potatoes):** Soaking sliced potatoes in hot water (140°F) for 10 minutes before baking activates pectinase enzymes, which break down cell walls and create a creamier texture without adding extra fat. The final dish has 25% less cream yet tastes equally rich.

**Secondary findings:**

**Tomato sauce:** A single base sauce (canned tomatoes + onion + butter) can be transformed into 6 different dishes by varying cooking time (15 minutes for fresh, 2 hours for concentrated), adding different herbs, or finishing with cream, wine, or anchovy paste.

**Pan material:** Cast iron retains heat 3–4 times longer than stainless steel (measured by temperature drop when cold food is added), making it superior for searing but worse for delicate sauces that need quick temperature changes.

**Salt timing:** Salting meat 45 minutes before cooking (not immediately before or 24 hours before) produces the juiciest results. The salt draws moisture out initially (first 10 minutes), then the moisture is reabsorbed with dissolved salt (after 30–45 minutes), seasoning the interior. Salting immediately before cooking only seasons the surface.

**Steak doneness:** Reverse sear reduces the temperature gradient from edge to center by about 7°F (from ~15°F to ~8°F). In practical terms, this means you can get a steak that is medium-rare from the edge to within 1/8 inch of the center, rather than having a 1/4-inch band of well-done meat around the outside.

**Turkey moisture:** Dry-brining saves approximately 15–20% of the meat's original weight in moisture. For a 12-pound turkey, that's about 1.5–2 pounds of water that stays in the meat instead of ending up in the pan. This translates to noticeably juicier meat — roughly equivalent to adding 2 tablespoons of butter per serving.

**Egg peeling:** Hot-start method reduces peeling time by about 70 seconds per egg (from 90 seconds to 20 seconds). For a batch of 6 eggs, that saves 7 minutes of frustrating peeling.

**Hollandaise stability:** Blender method reduces failure rate from ~30% (traditional whisking) to near 0% (blender method). The sauce also holds at room temperature for 2 hours without breaking, compared to 30 minutes for traditional hollandaise.

**Potato creaminess:** Hot-water soak reduces cream needed by 25% while maintaining the same perceived richness. For a standard 9x13 casserole, that's about 1 cup less heavy cream (saving ~800 calories).

**What the author acknowledges:**

These are kitchen experiments, not laboratory studies

Results may vary based on equipment, ingredient quality, and individual technique

Taste is subjective — what works for the test kitchen may not suit everyone's palate

The author has a background in food science (MIT graduate) but is not a professional researcher

**What a critical reader would note:**

**No peer review:** The book was not subjected to scientific peer review. Claims are based on the author's own testing.

**Small sample size:** Most taste tests involve 3–8 people, which is insufficient for statistical significance. A proper sensory panel would require 20–50 trained tasters.

**No blinding of the cook:** The author knows which condition is being tested, which can unconsciously influence cooking (e.g., paying more attention to the "experimental" steak).

**No replication:** Results have not been independently replicated by other labs or home cooks under controlled conditions.

**Publication bias:** The book only reports successful experiments. Failed experiments (where the conventional method worked better) are not included.

**Commercial interests:** The author runs a cooking website (Serious Eats) that sells cookbooks and kitchen equipment. There is potential bias toward methods that are more dramatic or surprising (and thus more marketable).

**Generalizability:** Test kitchen equipment (high-end stoves, professional-grade pans, digital thermometers) may not reflect typical home kitchens. Results may differ on electric vs. gas stoves, thin vs. thick pans, etc.

**No control for ingredient variability:** Eggs vary in size and freshness, meat varies in marbling and age, vegetables vary in water content. These are not controlled.

**No longitudinal data:** No testing of whether methods work consistently over multiple trials or across seasons.

For someone running their own n=1 experiment in the kitchen:

### What to test (specific intervention and dose)

**Reverse sear steak:** Cook steak in a 225°F oven until internal temp reaches 115°F (for medium-rare), then sear in a hot cast-iron pan for 60 seconds per side. Compare to traditional method: sear first in hot pan, then finish in oven.

**Dry-brined turkey:** Rub 1 tablespoon kosher salt per 5 pounds of turkey, refrigerate uncovered for 24–48 hours. Compare to wet-brined (1 cup salt per gallon water, 12–24 hours) and no-brine.

**Hot-start eggs:** Gently lower eggs into boiling water, boil 6 minutes for soft-boiled, 10 minutes for hard-boiled. Compare to cold-start (eggs in cold water, bring to boil, then time).

**Blender hollandaise:** 3 egg yolks + 1 tablespoon lemon juice in blender, slowly add 1/2 cup melted butter (120°F) while blending. Compare to traditional whisking method.

### Minimum meaningful duration

**Single-session experiments:** 1–2 hours per comparison (cook both methods simultaneously or back-to-back)

**For texture/moisture tests:** At least 3 repetitions on different days to account for ingredient variability

**For reheating tests (mac and cheese):** Test at 1 hour, 24 hours, 48 hours, and 72 hours after cooking

**For brining tests:** Allow 24–48 hours for dry-brining, 12–24 hours for wet-brining

### What to measure (specific metrics)

**Objective measures:**

- Internal temperature (use an instant-read thermometer, measure at center and 1/4 inch from edge)

- Weight before and after cooking (moisture loss percentage = (pre-weight - post-weight) / pre-weight × 100)

- Cooking time (minutes from start to finish)

- Crust thickness (cut steak in half, measure with ruler)

- Peeling time (seconds per egg, use a stopwatch)

**Subjective measures (use a 1–10 scale):**

- Tenderness (1 = tough, 10 = melts in mouth)

- Juiciness (1 = dry, 10 = very moist)

- Flavor intensity (1 = bland, 10 = very flavorful)

- Overall satisfaction (1 = would not eat again, 10 = best I've ever had)

**For sauces:** Measure time to break (minutes at room temperature before oil separates)

### Key confounds to control for

**Ingredient temperature:** Let meat come to room temperature (30 minutes out of fridge) before cooking. Use eggs straight from fridge for fair comparison.

**Pan temperature:** Use an infrared thermometer to ensure pans are at the same starting temperature. For searing, heat pan for exactly 5 minutes on medium-high.

**Oil type and amount:** Use the same oil (e.g., canola) and same amount (e.g., 1 tablespoon) for both conditions.

**Cutting technique:** Cut meat against the grain at the same angle. Slice eggs with the same knife.

**Resting time:** Let meat rest for exactly 5 minutes before cutting. Let eggs cool in ice water for exactly 5 minutes before peeling.

**Tasting order:** Taste samples in random order (flip a coin to decide which to taste first). Rinse mouth with water between samples.

**Blinding:** Have someone else label the samples (A, B, C) so you don't know which is which when tasting. This is the single most important control.

**Equipment:** Use identical pans, pots, and utensils for both conditions. If you only have one cast-iron pan, cook one condition, let pan cool, then cook the other.

### What a positive result would look like

**Reverse sear steak:** The reverse-seared steak has a temperature gradient of ≤8°F from edge to center (vs. ≥15°F for traditional), and you rate it 2+ points higher on overall satisfaction.

**Dry-brined turkey:** The dry-brined turkey loses ≤10% of its pre-cooking weight (vs. ≥20% for no brine), and you rate it 3+ points higher on juiciness.

**Hot-start