Biophysics · allometry · branching networks

WBE · The Geometry of Life

The left side abstracts vessels, airways, and plant vasculature into a space-filling delivery tree; the right places body mass and metabolism on logarithmic axes. Change organism, exponent, and terminal-unit assumptions to test why WBE predicts three quarters—and how that differs from a two-thirds surface-area hypothesis.

Body mass

M

$70.0\,\mathrm{kg}$

Whole-organism metabolism

B/B_1

$24.2\,\times$

Metabolism per mass

(B/M)/(B/M)_1

$0.35\,\times$

Characteristic time

\tau/\tau_1

$2.89\,\times$

Estimated terminal units

N_c

$2.01\times10^{10}$

Branch generations $N=21.6$

This is an interactive explanation of the idealized WBE model, not a claim that every species obeys exactly $\alpha=\frac34$ . Empirical slopes vary with taxon, temperature, activity, statistical method, and mass range; the original papers and representative challenges appear below.

Physics tutorial

Why do a shrew and a blue whale live at different tempos?

BackgroundMany biological quantities are not proportional to body mass $M$ , but approximately follow power laws. WBE treats vessels, airways, and plant vasculature as resource-delivery networks that repeatedly branch from one inlet until they service the complete organism.

Why it mattersIf total metabolism were controlled only by surface area, geometric intuition would suggest two thirds; classic compilations are often summarized by three quarters. WBE aims to explain metabolism, heart rate, circulation time, aortic radius, and capillary number with one network argument.

Start with the essentials

Focus question: How do space filling, invariant terminal units, and minimized transport cost turn a three-dimensional body into a three-quarter-power prediction—and which outputs change first when one assumption is relaxed?
One-sentence intuition: The key is not that a network merely looks tree-like. It must service the entire volume through approximately self-similar branching while keeping terminal exchange units size-invariant. Branching ratio changes the number of levels but cancels from the ideal exponent.

Core mathematical model

The common language of allometry

Y=Y_0M^b

Any trait $Y$ is assigned an exponent $b$ that states how rapidly it changes with body mass. On logarithmic axes, that exponent is the line slope.

WBE geometric compression

\gamma=n^{-1/3},\qquad \beta=n^{-1/2},\qquad \alpha=-\frac{\ln n}{\ln(\gamma\beta^2)}=\frac34

Space filling fixes the branch-length ratio $\gamma$ from daughter count $n$ ; area-preserving branching in the large-vessel regime fixes radius ratio $\beta$ . Substitution into the network-volume relation cancels $n$ and leaves three quarters.

One exponent resets the whole pace of life

B\propto M^{3/4},\qquad \frac{B}{M}\propto M^{-1/4},\qquad \tau\propto M^{1/4},\qquad N_c\propto M^{3/4}

Large organisms use more total energy but less per unit mass, run on longer characteristic times, and—if each terminal has invariant capacity—need terminal units in proportion to total metabolism.

Common difficulties

Fractal-like does not mean an infinite exact fractal

Typical misconceptionEvery vascular generation must be perfectly self-similar and branching must continue forever for WBE to work.

Better mental modelWBE uses finite, approximately self-similar delivery networks. Real vessels taper, branch asymmetrically, and cross from pulsatile to viscous flow. The theory claims those details are not the leading contribution across many orders of body mass.

Three quarters is not an uncontested natural constant

Typical misconceptionBecause WBE derives three quarters, every animal and plant dataset must land precisely on that slope.

Better mental modelA theoretical prediction and an empirical fit are different claims. Later analyses supported two thirds for some carefully selected mammal data and found heterogeneous exponents across vertebrate groups. The Lab therefore keeps an adjustable slope and both comparison lines.

Run the experiment

01
Compare the sizes of life
Restore WBE assumptions, then choose shrew, human, elephant, and blue whale while watching total metabolism, metabolism per mass, and characteristic time.
What to observe: Total metabolism rises rapidly but not as fast as mass itself. Metabolism per mass therefore falls while biological time stretches with a quarter-power exponent.
02
Turn the debate into a visible distance
Keep the blue whale selected and move the metabolic exponent slowly from $\frac23$ to $\frac34$ .
What to observe: The predictions nearly meet around one kilogram. Across seven or eight orders of body mass, their small slope difference becomes a large vertical gap—one reason dataset range matters so much.
03
Break terminal invariance on purpose
Raise terminal-size counterfactual exponent $\delta$ above zero, then switch between shrew and whale.
What to observe: Terminals now grow in large organisms, reducing the terminal count and branch generations required for the same total delivery. You changed a core WBE assumption, not a visual decoration.
04
Change daughters per branch
Move $n$ from two to four and compare network shape, radius ratio, and actual branch generations.
What to observe: A larger branching ratio reaches many terminals in fewer levels, yet the ideal WBE slope stays fixed because branching ratio cancels from the derivation.

WBE ideal model

\gamma=n^{-1/3},\quad \beta=n^{-1/2}

Three assumptions produce a family of predictions

A volume-serving, fractal-like network; size-invariant terminal units; and minimized transport cost jointly produce the quarter-power result. The animated tree is a representative slice, not a reconstruction of one species' vasculature.

Mechanistic prediction

B\propto M^{3/4}

The exponent links multiple traits

The important claim is not just a good metabolic fit. Whole-organism metabolism, metabolism per mass, biological times, vessel radii, and terminal counts should carry related exponents.

Empirical boundary

\alpha\approx\frac23\;\text{or}\;\frac34\;\text{or a varying slope}

The fitted slope is debated and context-dependent

Reanalyses have supported two-thirds scaling in selected mammal datasets, while broader comparisons report heterogeneous slopes across vertebrate groups. Use the comparison lines as competing hypotheses, not verdict and error bar.