How much of you is you?

A 2016 analysis worked from measured concentrations and volumes: most of the bacteria are in the colon, which holds about 0.4 litres, not the litre of gut the older estimate multiplied by. Human cells are dominated by red blood cells, which are numerous and small.

The result for a reference adult is roughly 38 trillion bacteria to 30 trillion human cells. A 2023 recount of human cells came to about 36 trillion, which leaves the two numbers about equal. The ratio varies between individuals and changes after a bowel movement, which says something about the precision available.

By mass the bacteria are a fraction of a kilogram. The imagery of being outnumbered and overwhelmed was doing a lot of work that the numbers do not support.

Gut bacteria break down complex carbohydrates the human gut cannot, some of them synthesise B vitamins, and colonisation shapes the adaptive immune system while it is developing. Mice raised germ-free develop abnormal immune systems, and they move more and behave less anxiously than mice carrying a normal gut community.

That much is demonstrated by intervention rather than inferred from correlation.

Before differences can mean anything, the normal range has to be known. A survey published in 2012 sampled 242 screened adults at up to eighteen body sites and sequenced what was there.

Its main finding was that the species present vary enormously between healthy people, while the metabolic functions those species perform are much more consistent.

That is part of why comparisons of species lists between sick and healthy groups have been hard to reproduce. Pooling ten obesity studies in 2016 found no consistent difference in the ratio of phyla everyone was quoting, and a classifier trained on one of those datasets was right about 57 per cent of the time on the rest. A 2024 analysis of 3,329 samples from seventeen countries reports a set of species that predicted obesity well in an independent dataset, so the signal may be real and small rather than absent. The stable quantity is what the community does, not who is in it.

Differences in gut communities are associated with obesity, inflammatory disease, allergy, and depression. Association is where most of these results stop.

Transplant experiments in mice sometimes transfer a trait along with a community, which is stronger evidence, and mice are not people. Germ-free mice given the gut community of an obese twin gained more fat than mice given her lean co-twin’s. One systematic review found that 36 of 38 published transfer studies reported the phenotype crossing into the animals, and argued that a success rate that high is not plausible and overstates what the microbiome does in human disease.

Faecal transplantation is an established treatment for one condition, recurrent Clostridium difficile infection, and that is the clearest clinical result the field has. One of the trials behind it was stopped early: 13 of 16 patients given donor faeces recovered after the first infusion, against 4 of 13 on vancomycin alone. For ulcerative colitis, trials suggest it may help, on evidence of low certainty.

Commercial probiotics are a different matter. Whether a strain settles in the gut at all varies from person to person, and in trials the effects depend on the strain and on the condition. In healthy adults, one review of trials found benefits for a few measures, stool consistency among them, and two found little sign of a lasting change to the gut community.