eduKateSG · How Science Works
Choose your reading route
Compare sizer, timer and adder predictions, follow original lineages and learn how single-cell evidence tests size control.
Define measurements and competing control rules · Chapters 1–3 · Start this route
Calculate size feedback and timing predictions · Chapters 4–6 · Start this route
Connect primary evidence with careful inference · Chapters 7–10 · Start this route
Full contents · Check your understanding · How Science Works master guide
Cells grow, divide and respond to changing conditions. Yet many populations of a particular cell type maintain a characteristic range of sizes rather than producing ever-smaller or ever-larger descendants. How can repeated growth and division correct size differences without requiring every cell to follow an identical schedule?
The answer begins by distinguishing a pattern from its mechanism. A population can show size homeostasis, meaning that deviations tend to be corrected, while individual cells vary substantially. Researchers follow cells through time and compare competing rules for when growth leads to division or another cell-cycle transition.
Three useful reference models are the sizer, timer and adder. They make different predictions under stated growth assumptions. None is a universal miniature ruler, clock or calculator that scientists have discovered inside every cell.
This lesson belongs beneath Cell Biology and the How Science Works master guide. Its worked examples are original, hypothetical calculations. They provide a way to interpret evidence without performing cell-culture or biological-intervention experiments.
Contents
Define measurements and competing control rules · Chapters 1–3
1. Decide what “size” means before comparing cells
Calculate size feedback and timing predictions · Chapters 4–6
4. Make the predictions numerically different
Connect primary evidence with careful inference · Chapters 7–10
7. Bacterial evidence links patterns with possible mechanisms
8. Other cells can coordinate growth and progression differently
9. Test the inference against measurement and sampling errors
CHAPTER 1 OF 10 · Define measurements and competing control rules
1. Decide what “size” means before comparing cells
Cell size can refer to length, projected area, surface area, volume or mass. Those quantities are related only through additional assumptions. A cell that becomes rounder may change its projected area without gaining the same proportion of biomass. A cell can also gain water without a matching increase in dry mass.
For a rod-shaped bacterium whose width remains approximately constant during elongation, length can be a useful size proxy. If width changes substantially, the relationship between length and volume changes too. The choice of proxy must therefore be checked against the conditions being studied.
A familiar geometry illustrates the issue. If a sphere’s radius doubles, its surface area increases fourfold and its volume eightfold. Saying that it is “twice as big” conceals which attribute changed. Real cells need not be spheres for this ambiguity to matter.
Define the measurement, units and cell-cycle landmarks together. “Volume just after birth” and “volume immediately before division” form a meaningful pair. A random snapshot of one young cell and one older cell does not reveal whether either has unusual size control.
The existing guide to comparable biological attributes develops the broader measurement-language problem. Here, we will use volume consistently and ask how its trajectory relates to cell-cycle progression.
CHAPTER 2 OF 10 · Define measurements and competing control rules
2. Track a lineage instead of trusting one average
Imagine a time-lapse diagram showing one cell at birth, during growth and just before division. Below it, show one daughter continuing the same sequence. Mark birth volume, division volume, elapsed time and the volume added between the two landmarks.
Write birth volume as Vb and division volume as Vd. The added volume is ΔV = Vd − Vb. Division time is another measured quantity, not something that can be inferred from the two volumes unless the growth trajectory is known.
A population average hides these relationships. Two populations could have the same average birth volume while differing in how quickly unusually small cells catch up. One might correct deviations in a single cycle, another gradually over several generations, and another maintain its average temporarily while its spread increases.
A lineage supplies successive observations, but it introduces dependence. A daughter inherits material and physiological state from its mother. Hundreds of generations from a small number of lineages do not automatically provide hundreds of independent biological replicates.
It also matters whether a dataset follows one continuing branch or samples all living cells at a moment in time. Faster-dividing branches can contribute more descendants to a population snapshot. Researchers must match their statistical comparison to how the cells were selected and followed.
The core question is conditional: for cells born at different sizes under comparable conditions, how do added volume, division volume and elapsed time differ? That question is more informative than asking whether the average cell “doubles”.
CHAPTER 3 OF 10 · Define measurements and competing control rules
3. Compare three deliberately idealised rules
For a sizer, imagine division occurring when volume reaches a fixed threshold. Small-born cells must add more volume than large-born cells to reach that threshold. In the ideal deterministic version, division volume is independent of birth volume.
For a timer, imagine a fixed interval from birth to division. Its size prediction depends on the growth law. If volume grows exponentially at the same specific rate in every cell, the interval multiplies each birth volume by the same factor. Large-born cells therefore remain larger at division.
For an adder, imagine a fixed volume increment between birth and division. A small-born and a large-born cell add the same amount in the idealised model, so their final volumes remain different. After equal division, however, the difference between their descendants is reduced.
These descriptions are reference cases. Experimental adder behaviour usually refers to a statistical pattern, such as the mean added volume being approximately independent of birth size over the tested range. It does not mean every cell adds an exactly identical amount.
The growth assumption is especially important for timers. If every cell instead grew linearly by the same absolute volume per unit time, a fixed interval would also produce a fixed added volume. The same apparent increment could therefore arise from a different combination of growth and timing rules.
A good model comparison states both the division rule and the growth law. Otherwise, a label can seem explanatory while leaving out the assumption that generated its prediction.
CHAPTER 4 OF 10 · Calculate size feedback and timing predictions
4. Make the predictions numerically different
Use three original hypothetical newborn volumes: 1, 2 and 3 arbitrary volume units. For the sizer, set the division threshold to 4. For the adder, set the added volume to 2. For the exponential-growth timer, choose an interval that doubles volume.
The sizer produces division volumes of 4, 4 and 4. Its added volumes are consequently 3, 2 and 1. A graph of added volume against birth volume slopes downward: each extra birth-volume unit removes one unit of required growth.
The adder produces division volumes of 3, 4 and 5. Its added volumes are 2, 2 and 2. The corresponding added-volume graph is horizontal, while division volume increases one-for-one with birth volume.
The timer produces division volumes of 2, 4 and 6. Its added volumes are 1, 2 and 3. In this particular doubling model, the added-volume graph slopes upward with slope one, and division volume increases with slope two.
Notice the shared middle observation. A newborn of volume 2 divides at volume 4 in all three examples. Studying only that typical cell cannot distinguish the models. The differences become visible when the predictions are tested across a range of starting sizes.
For a labelled diagram, place birth volume on the horizontal axis and added volume on the vertical axis. Draw the three predicted lines in separate colours, label the assumptions and plot the three synthetic points for each. Keep these model points separate from any experimental measurements. Agreement near one crossing point is weak evidence; agreement across a relevant range is more informative.
CHAPTER 5 OF 10 · Calculate size feedback and timing predictions
5. An adder can correct deviations over generations
The adder does not bring every cell to one identical division volume. How, then, can it stabilise size? Follow what happens after equal binary division, when each daughter receives half of the pre-division volume.
If a newborn has volume Vn and adds 2 units, the chosen daughter’s next birth volume is Vn+1 = (Vn + 2)/2. A stable birth volume is 2, because (2 + 2)/2 returns 2.
Start below that value at volume 1. Successive birth volumes are 1, 1.5, 1.75, 1.875 and 1.9375. Start above it at volume 3, and the sequence is 3, 2.5, 2.25, 2.125 and 2.0625. Both approach the same fixed point from opposite sides.
Subtract 2 from both sides of the recurrence. The deviation in the next generation is half the deviation in the current generation. This explains convergence without assuming that the cell directly measures how far it is from the fixed point.
Contrast the ideal doubling timer with equal partition. Its recurrence is Vn+1 = 2Vn/2 = Vn. A cell born at volume 1 produces a daughter born at volume 1, while a cell born at volume 3 produces one born at volume 3. Under these assumptions, the rule preserves the difference rather than correcting it.
Real growth, added size and partitioning fluctuate. The deterministic sequences demonstrate the direction of feedback, not a prediction of perfectly smooth biological trajectories. Noise creates variation around a characteristic range, and unequal division changes the recurrence. The simple model is useful precisely because its assumptions make the corrective effect visible.
CHAPTER 6 OF 10 · Calculate size feedback and timing predictions
6. Timing supplies a second test of the model
Assume exponential volume growth with a doubling time of one arbitrary time unit. Then V(t) = Vb × 2^t. The time needed to reach a specified division volume is t = log₂(Vd/Vb).
For the adder, the newborn volumes 1, 2 and 3 divide at 3, 4 and 5. Their model times are therefore log₂(3), log₂(2) and log₂(5/3): approximately 1.585, 1.000 and 0.737 time units.
The smallest newborn takes longest to add the same absolute volume because its initial absolute growth rate is smaller under exponential growth. Thus, equal increments do not imply equal cycle durations.
For the sizer threshold of 4, the same newborns require times of 2.000, 1.000 and approximately 0.415. For the doubling timer, all three take 1.000. Timing therefore supplies additional predictions even when one pair of starting and final sizes is shared.
These calculations would change if growth slowed during part of the cycle, if specific growth rates differed between cells or if the measured quantity were area rather than volume. Researchers should inspect growth trajectories rather than selecting a familiar equation because it makes the model easy to calculate.
A useful consistency check combines all three observations. Does a proposed rule explain the relationship between birth size and added size, the relationship with cycle duration, and the approach toward a stable range across generations? A model that matches one graph but contradicts the others needs refinement.
CHAPTER 7 OF 10 · Connect primary evidence with careful inference
7. Bacterial evidence links patterns with possible mechanisms
A 2015 study of bacterial size homeostasis used single-cell measurements to support an adder description in studied bacterial growth conditions. The work helped explain how size deviations can diminish through repeated growth and division without a pure fixed-size checkpoint.
The paper has a 2017 correction to an equation and a figure annotation; the authors state that the data and study results were not changed. The original arithmetic examples in this lesson do not use that corrected variability equation. Checking corrections is part of following evidence, even when a central conclusion remains intact.
A later 2019 experimental study investigated how balanced production of division-related components and their accumulation toward a threshold could generate adder behaviour. Perturbing division-protein dynamics and replication-initiation dynamics produced different effects in the studied Escherichia coli and Bacillus subtilis systems.
The conceptual advance is the move from describing a statistical pattern to perturbing candidate causal machinery. If an intervention changes one proposed control process but leaves the predicted size behaviour intact, that result constrains the explanation. If another intervention changes the behaviour as predicted, it supplies a different kind of support.
This does not make one molecular account universal across bacteria, growth conditions or cell types. Even the studied systems can show deviations under particular conditions. “Adder” is a compact description of a relationship; identifying the molecules and events responsible requires further experiments.
CHAPTER 8 OF 10 · Connect primary evidence with careful inference
8. Other cells can coordinate growth and progression differently
A cell cycle contains multiple transitions, so control can act at more than one point. A cell might spend longer before a particular transition when it is small, alter its growth rate, or combine several responses. Its total birth-to-division pattern can reflect those processes together.
A 2018 study of mammalian cells examined contributions from both growth-rate regulation and cell-cycle duration. This broadens the question beyond whether division waits for one threshold: the trajectory leading toward division can itself be regulated.
More recently, a 2025 study of epidermal stem cells in living tissue found size-dependent control of entry into S phase involving the RB pathway in the studied mouse system. Features of the surrounding microenvironment influenced growth, while the coupling between current size and the transition showed cell-autonomous regulation.
That result concerns a particular transition and experimental context. S-phase entry is not the same event as completed cell division, and evidence from epidermal stem cells should not be silently extended to every mammalian cell. The distinction helps explain why different studies can report different effective size-control patterns without simply contradicting one another.
Size homeostasis also does not mean every cell type shares one target size. Differentiated cells perform different jobs and may stop dividing or grow in unusual ways. The useful comparison is within a defined cell type, state and environment, with a clear statement of which transition or whole-cycle behaviour is being explained.
CHAPTER 9 OF 10 · Connect primary evidence with careful inference
9. Test the inference against measurement and sampling errors
A graph of added volume against birth volume appears simple, but the same birth measurement occurs on both axes. Added volume is calculated by subtracting birth volume from division volume. An error that makes birth volume look too large simultaneously makes the calculated added volume look too small.
This shared-error structure can create an artificial negative relationship. It can make a dataset look more sizer-like even if the underlying cells do not follow that rule. Estimating segmentation uncertainty and testing its effects are therefore part of model comparison, not optional polishing after a line has been fitted.
Changing cell shape introduces another possible error. If length is used as a proxy while width changes with birth size or condition, the apparent slope can reflect geometry as well as regulation. Independent checks of the size metric help separate those contributions.
Environmental mixtures can also mislead. Combining slow- and fast-growing conditions may create a population trend that differs from the relationship within either condition. Compare appropriately grouped observations and state whether a model describes within-condition variation, changes between conditions or both.
An original paper worksheet can use the three synthetic newborns from Chapter 4. First calculate division volume, added volume and time for each rule. Then mark the shared point at birth volume 2. Finally, ask which additional birth sizes and which independent measurements would most clearly distinguish the predictions.
The worksheet teaches a general principle: deliberately inspect where explanations disagree. Collecting more observations at a point where all models predict the same outcome may improve precision without improving discrimination. A well-chosen comparison can be more informative than a large undifferentiated average.
CHAPTER 10 OF 10 · Connect primary evidence with careful inference
10. Check the feedback logic rather than memorising labels
What is being stabilised? A characteristic size range within defined conditions, not an identical measurement in every individual. Changes in cell type, environment or developmental state can change that range.
Why does the simple adder restore the birth volume toward 2? Adding 2 and then dividing equally halves the previous deviation from 2. Small-born and large-born lineages therefore converge in the deterministic model.
Why does the doubling timer fail to correct deviations in this example? Doubling followed by halving returns each lineage to its own previous birth volume. There is no restoring effect under those stated assumptions.
Can a timer ever look like an adder? Yes. With the same linear absolute growth rate, a fixed elapsed time produces a fixed added amount. That is why growth laws belong beside model labels.
Does an adder-like graph identify a molecular mechanism? No. It describes a statistical relationship. Molecular explanations require additional evidence, including measurements and perturbations that distinguish plausible causal routes.
The scientific achievement is to connect growth, timing, partitioning and evidence into a testable account. Follow the Cell Biology guide for related control systems, or return to the How Science Works master to compare how models explain other natural patterns.