The test kitchen
How this site will test its own kitchen advice, and the experiments designed so far. None of them has been run.
Nothing on this page is a result
Every experiment below is a plan. No measurement has been taken, no tasting held and no photograph made. Each is marked Proposed — not yet performed, and will stay that way until it has been carried out under the rules here and its raw measurements are published with it. The hypotheses are predictions written in advance, not findings.
Why test at all
Most kitchen advice about chocolate is inherited. It is repeated because it is plausible and because the last person said it, and a good deal of it has never been checked by anyone who wrote it down. The rest of this site reports what is known and how well; this part is for the questions where the honest answer is that somebody should simply try it, several times, and write down what happened.
The tests are kitchen tests. They can show that one method gave a glossier bar than another in these hands with this chocolate. They cannot replace a food scientist’s instruments, and nothing here will be offered as a validated safety procedure.
The rules
- The question and the hypothesis are written down before the test.A hypothesis written afterwards always fits. Committing the design first is what makes a surprising result believable and an expected one worth anything.
- One thing is changed at a time.If the chocolate and the temperature both differ between two cups, the difference between the cups belongs to neither.
- Every condition is run at least three times, on more than one day.Chocolate work varies from run to run for reasons that have nothing to do with what is being tested. One run of each condition measures that variation and calls it a result.
- What can be measured is measured, with a named instrument.Temperature, mass and time are cheap to measure properly. 'Thick' and 'glossy' are not measurements, and where a judgement has to be used it is made blind and by more than one person.
- Tasting is blind.Anyone who knows which sample is which will taste what they expect. Samples are coded by someone who does not taste.
- The products used are named, with their batch codes.A result for 'dark chocolate' is a result for one bar. Naming it lets a reader judge how far it travels, and lets anyone repeat it.
- Raw measurements are published with the result.A summary cannot be checked. The readings, including the runs that went wrong, are committed beside the write-up.
- Failed and inconclusive experiments are published too.An archive of experiments that all worked is evidence of selection, not of anything about chocolate.
- Photographs are of the actual runs.No stock imagery, no illustration standing in for a result. If it was not photographed during the test, there is no photograph.
- A result is a result for a domestic kitchen.Nothing here is a validated food-safety procedure or an industrial trial, and no result is offered as one.
What a published experiment will contain
The same parts every time, so that two experiments can be compared and so that a missing part is obvious: the question; the hypothesis as written beforehand; what was changed and what was held constant; the products, with batch codes; the equipment; every measurement, with the instrument; the number of runs; the raw readings; photographs of the runs; what went wrong; the result; and what the design could not have shown.
The record format is enforced in the site’s code: an experiment cannot be marked as performed unless its date, its repeats, its equipment, its photographs and a file of raw measurements are all present, and the site will not build if they are not.
Designed experiments
Proposed — not yet performedDrinks
Hot chocolate from chocolate, from cocoa powder, and from both
- Question
- Does hot chocolate made from a bar actually taste richer than one made from cocoa powder, at the same cocoa content?
- Prediction
- At matched cocoa solids and sugar, tasters will rate the bar-based drink as fuller-bodied, because it carries more cocoa butter, and will not reliably tell them apart on chocolate intensity.
- Changed
- Source of cocoa: bar only, cocoa powder only, or half of each.
- Held constant
- Total cocoa solids and total sugar per serving, calculated from the labels; The same milk, from the same carton; Serving temperature; Whisking time and vessel.
- Measured
- Serving temperature: Probe thermometer, to 1°C, at the moment of serving.
- Body, intensity, sweetness and preference: Blind ranking by at least four tasters.
- Time to a visible skin or sediment: Timed, from serving.
- Runs
- At least 3 of each condition.
- Tests the advice in
- Making hot chocolate
Cannot show: Anything about a different bar or a different cocoa powder, or about a drink made with water.
Proposed — not yet performedDrinks
What actually thickens hot chocolate
- Question
- How much of a thick hot chocolate's body comes from the chocolate, and how much from starch?
- Prediction
- Doubling the chocolate will thicken the drink less than a small addition of cornflour, and the two will be distinguishable blind by texture.
- Changed
- Chocolate per serving, at three levels; Cornflour: none, or one fixed addition.
- Held constant
- The same chocolate and milk; Heating method and final temperature; Measurement temperature.
- Measured
- Flow: Time for a fixed volume to drain through a funnel, in seconds, at a stated temperature.
- Perceived thickness: Blind ranking by at least four tasters.
- Runs
- At least 3 of each condition.
- Tests the advice in
- Hot chocolate calculator
Cannot show: A viscosity in scientific units. A funnel time compares these drinks with each other and nothing else.
Proposed — not yet performedMelting
Microwave or bain-marie
- Question
- Is melting chocolate in a microwave really more likely to overheat it than a bain-marie?
- Prediction
- The microwave will produce a higher peak temperature and a wider spread between the hottest and coolest points, and that spread, not the method, will predict graininess.
- Changed
- Method: microwave in fixed bursts, or bowl over hot water; Form: chopped bar or chips.
- Held constant
- Mass of chocolate; The same chocolate, same batch; Bowl material and size; Stirring schedule.
- Measured
- Temperature at the centre and the edge: Probe thermometer, to 1°C, at each stir.
- Time to fully melted: Timed.
- Texture when cooled: Blind assessment for graininess by at least three people.
- Runs
- At least 3 of each condition.
- Tests the advice in
- Melting chocolate
Cannot show: How a different microwave, power setting or bowl behaves.
Proposed — not yet performedTempering
How much seed chocolate tempering needs
- Question
- Does it matter how much unmelted chocolate is used to seed a batch?
- Prediction
- Below a certain proportion of seed the chocolate will set dull or streaked however carefully the temperatures are followed, and above it more seed will make no visible difference.
- Changed
- Seed as a share of the batch, at four levels.
- Held constant
- The same couverture, same batch; Melt temperature and working temperature; Room temperature and humidity, recorded; Mould and cooling method.
- Measured
- Temperature through the process: Probe thermometer, to 0.5°C, logged every minute.
- Set time on a test strip: Timed at a recorded room temperature.
- Gloss, streaking and snap: Blind scoring against photographed reference pieces.
- Bloom after four weeks: Inspection and photograph of stored pieces.
- Runs
- At least 4 of each condition.
- Tests the advice in
- Tempering chocolate
Cannot show: The crystal form present. That needs an instrument a kitchen does not have; this measures the appearance that form produces.
Proposed — not yet performedGanache
Ganache ratio and how firmly it sets
- Question
- How well do the standard chocolate-to-cream ratios predict a ganache's firmness, across dark, milk and white?
- Prediction
- Firmness will track the ratio within one type of chocolate but not across types: a milk ganache at the dark ratio will set markedly softer.
- Changed
- Ratio of chocolate to cream, at three levels; Chocolate: dark, milk, white.
- Held constant
- The same cream, same fat content; Method and emulsifying time; Setting temperature and time; Container.
- Measured
- Firmness: Depth a fixed weight sinks in a fixed time, in millimetres.
- Whether it split: Recorded yes or no, with a photograph.
- Runs
- At least 3 of each condition.
- Tests the advice in
- Ganache calculator
Cannot show: Shelf life or safety of the ganache. Those depend on water activity, which is not measured here.
Proposed — not yet performedStorage
Cupboard, fridge and freezer
- Question
- Does refrigerating chocolate damage it, or is the damage done when it comes out?
- Prediction
- Chocolate stored cold and brought back to room temperature in its wrapping will be indistinguishable from cupboard-stored chocolate; unwrapped cold, it will develop sugar bloom.
- Changed
- Storage: cupboard, fridge, freezer; Return to room temperature: wrapped, or unwrapped.
- Held constant
- The same bars, same batch; Storage time; Wrapping; Room conditions at unwrapping, recorded.
- Measured
- Surface condition: Photograph under fixed lighting at set intervals, scored blind.
- Room temperature and humidity: Thermo-hygrometer, logged.
- Taste and texture: Blind triangle test against the cupboard sample.
- Runs
- At least 3 of each condition.
- Tests the advice in
- Storage advisor
Cannot show: What happens over months, or to filled chocolates.
Proposed — not yet performedBaking
Natural or Dutch-processed cocoa in a cake
- Question
- What does swapping one cocoa for the other actually change in a baked cake?
- Prediction
- In a recipe raised with bicarbonate of soda, Dutch-processed cocoa will give a darker, flatter cake, and the difference in rise will be measurable.
- Changed
- Cocoa: natural, or Dutch-processed; Raising agent: bicarbonate of soda, or baking powder.
- Held constant
- Every other ingredient, weighed; Tin, oven position and oven temperature, checked with an oven thermometer; Bake time.
- Measured
- Height at the centre: Ruler, to 1 mm, after cooling.
- Colour: Photograph beside a grey card under fixed lighting.
- Flavour and preference: Blind ranking by at least four tasters.
- Runs
- At least 3 of each condition.
- Tests the advice in
- Natural versus Dutch-processed cocoa
Cannot show: The result in a different recipe. Acidity matters differently depending on what else is in the batter.
Proposed — not yet performedBaking
Chips or a chopped bar in a cookie
- Question
- Do chocolate chips really hold their shape better than a chopped bar, and does anyone prefer one?
- Prediction
- Chips will keep a distinct shape and chopped chocolate will spread into the dough; preference will split rather than favour either.
- Changed
- Chocolate form: chips, or a bar chopped to a similar size.
- Held constant
- Cocoa percentage of both, from the labels; Dough, weighed and rested the same time; Oven temperature and bake time.
- Measured
- Whether pieces kept their shape: Photograph of a cut cookie, scored blind.
- Preference: Blind paired comparison by at least six tasters.
- Runs
- At least 3 of each condition.
- Tests the advice in
- Baking with chocolate
Cannot show: Whether the difference is the shape or the recipe, since chips and bars are formulated differently. The labels are recorded so that can at least be seen.
All research areas