Research
What individual studies of cacao and chocolate actually asked, did and found — and, for each, what it does not show.
Cacao genetics and breeding
Ten genetic clusters where tradition saw three types; cacao in use 5,300 years ago; and how far genomic breeding has actually got.
14 studies
Cacao fermentation
Which microbes do the work, which of them are necessary, and how much of the textbook account rests on a handful of heaps.
26 studies
Cacao pollination
Whether pollination, rather than fertiliser, is what holds cocoa yields down, and who the pollinators really are.
11 studies
Salmonella and chocolate
The outbreaks, how long Salmonella survives in dry food, and what roasting, conching and dry cleaning have been measured to do.
16 studies
Allergens in chocolate
How much milk and peanut turns up in chocolate that does not list them, and what a “may contain” label has been measured to predict.
8 studies
The science of making chocolate
How chocolate flows, what conching and tempering physically change, and where bloom comes from.
20 studies
Shelf life and storage
What storage trials found about bloom, migrating fat, spoilage in fillings and rancid nuts, and why none of them yields a general shelf life.
22 studies
Packaging and migration
What has been measured moving from sacks, board, inks and plastic into cocoa and chocolate, and what life cycle assessments say about packaging’s share.
13 studies
Cadmium in cacao
Why cadmium varies farm by farm, what the tree and fermentation do with it, and why no soil treatment has yet been shown to lower it in beans.
19 studies
Lead, mould toxins and other contaminants
Which contaminants sit in the shell and leave with it, which are made by roasting, and how much the choice of benchmark changes the headline.
18 studies
Farm incomes, labour and forests
What has been measured on cocoa farms and from satellites: income, yield, child labour, forest loss, and what certification does and does not coincide with.
38 studies
Cocoa agroforestry
Whether shade costs yield, what it stores and shelters, and why the answer differs between Indonesia, Bolivia and Cameroon.
10 studies
Cacao and climate
What models project for West Africa, the Amazon and Mesoamerica, what drought and heat have been measured doing, and why shade does not reliably protect.
24 studies
Cacao pests and diseases
Where each disease came from, what it has been measured to cost, and what fungicide, sanitation, biocontrol, barriers and resistance achieved in trials.
38 studies
Cacao by-products
What has been done with pod husk, pulp and shell, and at what scale: almost all of it in a laboratory.
16 studies
Roasting, refining and alkalising
What different roasters, mills and alkali treatments have been measured doing to aroma, polyphenols and flow.
17 studies
Chocolate and health: the trials
What the largest trial and the systematic reviews found, and why most of it is about flavanol extracts and not chocolate bars.
26 studies
Who eats chocolate, and how much
What dietary surveys show about how many people eat chocolate on a given day, and how much.
7 studies
Cacao biotechnology and biological control
Gene editing, tissue-culture propagation and microbes living inside the tree: what has been shown in a laboratory and what has not been shown in a field.
5 studies
From flower to flavour
The studies above arranged as one chain, from pollination to roasting, with what has been measured at each link and what has not been shown.
Nine steps, across three of the areas
Our own testing
The studies above are other people’s work. For the kitchen questions nobody has published an answer to, this site has written down how it will test them itself, and what it will test first. No experiment has been run yet, and the page says so: the test kitchen.
How a study is summarised here
- Question
- What the researchers set out to find, in one sentence.
- How
- The design and the sample: how many trees, heaps or cases, where, and by what method.
- Finding
- The main result, with its figure where the paper gives one.
- Does not establish
- What a reader might take from the headline that the study does not support. This is the reason the section exists.
- Limits
- What about the design restricts how far the finding travels: one site, laboratory scale, one strain.
- On this site
- The pages where the subject is explained in full.
How far each study was read
Every study here was read in its published abstract, and a phrase from each entry was checked to appear there. Where a paper is open access, more of it was read. That is a real limit and it is stated plainly: an abstract is the authors’ own summary, and it usually states a result more firmly than the paper does. So the limits given for each study are the ones visible in its design, and where an abstract does not give a sample size, the entry says so instead of supplying one.
Nothing here comes from a press release or a news report about a study. Studies are chosen because they are the ones a claim about chocolate usually traces back to, not because they agree with each other, and several of them do not.
For how evidence is graded on health pages, which is a separate and stricter exercise, see sources and evidence.