Last time I promised you a glass tube of pond mud that sorts itself into coloured bands and then holds them. It does — but the colours turn out to be the least reliable part of the arrangement.
The recipe has barely changed in a century. Take sediment from the edge of a pond. Mix in a carbon source — shredded newspaper, leaf litter, chopped lettuce — and a sulfate source, usually gypsum. Pack it into a clear column, pressing hard to force out trapped air. Top it with a few centimetres of pond water, cover it, and stand it in the light.
Then never feed it again.
Within weeks the grey uniformity is gone. A green skin forms at the waterline. The bottom goes black. In between, if you are lucky, the glass develops stripes. Left alone it holds that arrangement for months, often years.
Nothing sorted the mud but the mud
What does the sorting is two gradients running through the tube in opposite directions. Oxygen enters only at the top and is consumed faster than it diffuses, so it runs out a short distance below the waterline. Everything under that line is permanently anoxic.
Down there, bacteria that cannot use oxygen respire sulfate instead — the gypsum you stirred in — and the waste is hydrogen sulfide, which diffuses upward. Where it meets iron it precipitates as ferrous sulfide, and that is what turns the bottom black. Where it meets light before it meets oxygen, it becomes food: photosynthetic bacteria strip electrons off it to build sugar, exactly as a plant strips them off water, and hand back sulfur and sulfate, which sink. The same atoms go up and down the tube indefinitely.
The sulfide was not in the jar when you sealed it. The gradient that decides where every organism lives is manufactured, continuously, by those same organisms. Each one’s address is the depth at which its chemistry balances between a poison falling from above and a meal rising from below — and both are excreted by neighbours. The column is not a set of shelves that life moved into. It is a map its own residents are drawing, and being filed by, at the same time.
That is the distinction owed to No. 3, the bottle garden sealed in 1960 and watered once since. Both are closed to matter and paid only in photons. But the bottle garden is one world, mixed; its loop runs in time, oxygen made by day and spent by night. This one runs in space. It does not merely cycle — it sorts, and keeps the arrangement. Nor is it No. 4’s sourdough starter, which is fed, or No. 13’s lichen, a bargain between two partners. This is a whole town, and nobody is delivering anything.
The stripes are optional
The canonical diagram puts a red-violet band of purple sulfur bacteria in the upper middle and a green band of green sulfur bacteria below it. In 2014 a group sequenced twelve columns — mud from two ponds near Williams College, two carbon sources, eighteen weeks under controlled light — and found the purple sulfur bacteria and their purple non-sulfur relatives “were not found to be abundant in these columns.” The green sulfur bacteria were there, but scattered “rather than in the expected lower-middle zone.” Their flat summary: distinct green and red-violet zones “were not apparent,” only patches of colour at varying intensity.
And yet the layering was emphatic — just written in DNA instead of pigment. Proteobacteria were most abundant at the top and thinned with depth; Bacteroidetes and Firmicutes did the reverse. Cyanobacteria marked the soil-water interface, Clostridia the deepest samples. At eight centimetres they found methane-producing archaea, and four centimetres above them, methane-eating bacteria living on what was rising from below. The authors note honestly that they sampled by drilling into the column’s side, which “may have excluded or destroyed surface-attached members” — some colour may have been clinging to glass they could not reach.
Still: the structure did not need the stripes. The photograph everyone reproduces is the pattern’s most photogenic possible outcome, not its definition.
The job is not the worker
The second finding is stranger. Both ponds’ columns ran a functioning sulfur cycle, and ran it with unrelated organisms. In columns from one pond the sulfate reduction was done almost entirely by Deltaproteobacteria; in the other, by a family of Firmicutes — a different phylum, an enormous evolutionary distance away. The paper puts the general case plainly: sulfur-reducing bacteria “are not a monophyletic group, but rather are defined physiologically.” They are not a family. They are a job description, and the column hires locally.
Which leaves a strange accounting. The atoms cycle and are conserved. The cells are born and die continuously. The species are contingent — a different bucket of mud staffs the same structure with different creatures. What persists, for years, is a set of chemical occupations and the distances between them.
There is a last irony in the name on it. Sergei Winogradsky’s great discovery, made by watching sulfur granules appear and vanish inside Beggiatoa, was that some organisms build themselves from inorganic chemistry alone, with no light anywhere in the process. He coined the word chemosynthesis to mark it off from photosynthesis. The apparatus that carries his name runs on nothing else.
≈
One loop I’m watching
Next: a structure inside your cells that holds a fixed length while every molecule in it is marching. Subunits lock on at one end and fall off the other at exactly the same rate, so the filament stays put while its substance travels end to end and leaves — a standing wave a few millionths of a metre long, and the reason a crawling cell can move without anything inside it going anywhere. Next time.