How bacteria grow, and why counting them is hard

When a microbiologist says a culture is growing, they mean the population is doubling, not that individual cells are getting larger. Each cell splits into two, both daughters split again, and the count follows one equation:

$N(t) = N_0 \cdot 2^{t/g}$

Here $N_0$ is the starting count, $t$ is elapsed time and $g$ is the generation time, the interval a population needs to double. For Escherichia coli in rich medium at 37 degrees Celsius, $g$ is about 20 minutes. In a nutrient-poor environment the same species may take days.

Generation time is a property of the conditions as much as of the organism. Change the temperature, the carbon source or the oxygen, and $g$ moves.

Start with a single cell and let it double every twenty minutes. Nothing dramatic seems to happen for the first hour.

After one hour there are 8 cells. After two, 64. After three, 512. The curve looks flat and then looks vertical, but the rule never changed: it multiplied by two, ten times over. This is why lab cultures are plotted on a logarithmic axis, where exponential growth becomes a straight line and a change in slope becomes visible.

Extend that curve and it stops making physical sense very quickly.

The point is not the absurd number. It is that unrestricted exponential growth is never what you are observing. Any real culture is already being limited by something, and identifying that limit is most of what growth experiments are for.

Inoculate a sealed flask of broth and the population traces the same four-part shape every time. Nothing is added and nothing is removed, which is what makes the pattern reproducible.

Lag is not idleness. A cell moved from one sugar to another has to build a different set of enzymes before it can divide, and the length of the lag measures that retooling.

Each transition is triggered by a change in the cells' surroundings rather than by a clock, which is why altering the medium moves the boundaries.

You cannot count a billion cells individually, so every method is a proxy, and the proxies disagree on purpose.

Key idea: Optical density answers "how much bacterial material is here". A plate count answers "how much of it can reproduce". After a disinfectant or an antibiotic these two numbers separate sharply, and which one you quote decides what you conclude.

A viable count works by dilution. You dilute the sample until a plate carries a countable number of colonies, conventionally 30 to 300, then scale back up. Below 30 the count is dominated by chance; above 300 colonies merge and hide each other.

def cfu_per_ml(colonies, volume_plated_ml, dilution_factor):
    """Scale one countable plate back to cells per mL of the original sample."""
    return colonies / (volume_plated_ml * dilution_factor)

# 137 colonies from 0.1 mL of a one-in-a-million dilution
cfu_per_ml(137, 0.1, 1e-6)
# 1.37e+09 CFU per mL

So 137 countable colonies report 1.37 billion viable cells per millilitre in the original tube. Every plated cell is assumed to have produced exactly one colony, which is where the method starts to lie.

The name is a confession. A plate count does not measure cells, it measures things that grew into a visible colony.

Gotcha: Many bacteria grow in pairs, chains, or clumps. A clump of forty staphylococci that lands on the agar together produces one colony and is recorded as one CFU. The same sample viewed under a microscope shows forty cells. Neither number is wrong; they are answers to different questions.

This is why Staphylococcus samples are vortexed before plating and why counts are reported as CFU rather than as cells. It also means a plate count is systematically an underestimate of cell number whenever the organism aggregates, which is most of the time outside a well-shaken flask.

There is a deeper problem than clumping. Cells can be intact, metabolising and capable of causing infection, yet produce no colony on any plate you offer them. Microbiologists call this state viable but nonculturable, or VBNC.

Bacteria enter it under stress: starvation, cold, low oxygen, disinfectant residue. Membranes stay intact and genes stay transcribed, but division stops. On a plate count the sample reads clean. Under a microscope with a viability dye, it is populated.

In practice: This is why food and water safety standards specify the exact medium, temperature and incubation time, and why a negative plate count is evidence of absence only for organisms that grow under those precise conditions.

Counting is not bookkeeping. Nearly every practical question in microbiology is a comparison of two counts.

Does this disinfectant work? Compare viable counts before and after. Is this water safe? Count indicator organisms per 100 mL against a legal threshold. Does this antibiotic stop this isolate? That is the next lesson, and it is a growth measurement wearing different clothes: the lowest drug concentration at which the population fails to increase.

The recurring trap is treating one proxy as the truth. Optical density that will not fall does not mean the drug failed, and a clean plate does not mean the sample is sterile. Know which question your number answers.