how-to · 5 min read
Cell doubling time and growth curves, explained
A growth curve describes how a population of cells changes over time, and doubling time describes how quickly it doubles while growing exponentially. Together they are the standard language for comparing bacterial, yeast or cultured cell growth, but only when the measurement comes from the part of the curve where exponential growth actually holds.
This guide explains what doubling time and generation time mean, works through a calculation from two counts, shows how to read the phases of a growth curve, and lists the assumptions and mistakes that make a reported doubling time misleading.
Doubling time and generation time
Doubling time, written T_d, is the time a population needs to double in size under balanced exponential growth, following N(t) = N₀ · 2^(t/T_d). Generation time is the time between one cell division and the next. In a balanced culture the two are numerically the same, because one doubling corresponds to one generation on average.
The exponential growth rate µ = ln2/T_d expresses the same information as a per-capita rate per unit time. A culture with a 2-hour doubling time has µ ≈ 0.3466 h⁻¹, while a 4-hour doubling time gives µ ≈ 0.1733 h⁻¹. Doubling time and growth rate are inversely related, so either one can describe a condition once the units are stated.
A worked example
Suppose a culture grows from 1.0×10⁵ cells per mL to 8.0×10⁵ cells per mL in 6 hours. The count increased by a factor of 8, and because 2³ = 8 the population doubled three times. The generation number is n = log2(8) = 3. Dividing the elapsed time by the number of doublings gives T_d = 6 h / 3 = 2.0 h.
The same numbers via the formula give T_d = t · ln2 / ln(N_t/N₀) = 6 × 0.6931 / ln(8) = 6 × 0.6931 / 2.0794 = 2.0 h. The growth rate is µ = ln2 / 2.0 h = 0.3466 h⁻¹. Both answers describe the same exponential phase; the generation count is useful for intuition, and the rate is useful for modelling and comparison.
Reading the phases of a growth curve
A batch culture typically passes through lag, exponential, stationary and death phases. In lag phase the cells adapt and the count may barely rise; during exponential phase the count increases by a constant factor per unit time; in stationary phase growth and death balance; and in death phase the viable count falls. A doubling time computed across lag or stationary phase mixes rates and describes no single phase.
Plotting counts on a logarithmic axis makes exponential phase look like a straight line, because log(N) = log(N₀) + (µ/ln10) · t. The slope gives the rate directly, and a straight stretch is the visual signal that exponential growth holds. A curve that bends upward or downward is telling you that the growth rate is changing, so a single pair of counts should be treated with caution.
Choosing between doubling time and growth rate
Doubling time is intuitive: it answers how long until the population doubles, and it makes differences between conditions easy to see. Growth rate is more convenient in equations, particularly when fitting exponential models or comparing with other rate constants, and it is the natural parameter when counts are plotted on a log scale.
The two are linked by µ = ln2/T_d, so reporting both is harmless and often helpful. What matters is specifying the unit: 2 h, 120 min and 0.0833 d are the same doubling time, but a growth rate of 0.3466 h⁻¹ is not the same as 0.3466 min⁻¹. Keep the counts and times in one consistent unit before comparing conditions.
Assumptions, errors and pitfalls
A clean doubling time assumes balanced exponential growth with a constant per-capita rate, no lag or death phase, and no nutrient limitation or density-dependent slowing. It also assumes that every time point measures the same quantity the same way: total particle counts and viable-cell counts are different measurements, and mixing them across a series will distort the result.
Practical sources of error include counting noise, sampling from a non-homogeneous culture, clumping, dilution mistakes and an optical density reading that has not been converted to cells with care. If precision matters, take several time points during exponential phase, plot log count against time, fit a straight line and report the slope with its uncertainty rather than deriving a rate from one pair of counts.