C.H. Waddington

An Embryologist Who Read Too Widely

Conrad Hal Waddington trained first in geology at Cambridge, then turned to embryology under the influence of Alfred North Whitehead's process philosophy — a debt he never hid and that shows up everywhere in his later work as an instinct for describing organisms as unfolding processes rather than assembled parts. By the 1930s he was working on induction in chick and amphibian embryos, the phenomenon by which one tissue instructs an adjacent one to differentiate. What he kept noticing, across species and across perturbations, was not fragility but the opposite: embryos reached the same endpoint through visibly different routes, and even genetic or environmental insults that clearly altered the process along the way were very often absorbed without altering the outcome. Development, whatever it was, resisted being knocked off course.

In 1942 he gave that resistance a name in a short Nature paper, "Canalization of Development and the Inheritance of Acquired Characters." A developing organism, he proposed, follows a canalized pathway — buffered against small variation in genotype and environment alike, so that the visible phenotype stays constant even while the underlying causal factors vary. The paper is three pages long and contains no differential equation. It did not need one to be right; canalization as an empirical phenomenon was, and remains, well documented independently of any particular mathematical model of it.

The Epigenetic Landscape

Fifteen years later, in The Strategy of the Genes (1957), Waddington gave canalization its lasting image. He commissioned the artist John Piper to draw a landscape of branching valleys sloping down from a ridge. A ball — the developing cell — sits at the top and rolls downhill. Where the ball ends up is its fate: which cell type it becomes. The valley floor is not flat but ribbed, so that a ball nudged sideways rolls back to the valley's centerline rather than drifting into the wrong valley — this is canalization, drawn rather than proved. Where a valley forks, a small difference in position can send the ball down one branch or the other; Waddington called these branch points chreodes (from the Greek for "necessary path"), and the underlying pegs and guy-ropes he sketched beneath the landscape, pulling the fabric of the valley walls into shape, were his visual shorthand for the genes and their interactions actually doing the work.

It is one of the most reproduced diagrams in twentieth-century biology, and almost none of the reproductions show what Waddington drew underneath it: a tensegrity structure of pegs in the ground, connected by guy-ropes to the sheet forming the landscape's surface, each rope representing a gene's pull on the developmental terrain. He was explicit that the landscape was a metaphor for a real, gene-determined topography, not decoration — but he never wrote the equation that the topography was a picture of. In 1957 the mathematics to do so properly barely existed; Thom's own classification theorem was still fifteen years from publication.

The Landscape, Formalized

This corpus's own opening note to WP-30, "The Missing Anchor", invoked Waddington's landscape directly, calling it "a curvature picture… drawn by a biologist who did not know that is what he was drawing." That line was not incidental. The technical treatment this series gives the landscape lives in Book 6, "Cell Fate as a Lyapunov Attractor", which writes the ball-in-valley picture as a genuine potential $V(x)$ with cell fate identified as a local minimum, canalization as the negative-definiteness of the Hessian at that minimum (Lyapunov stability, not metaphor), and chreodes as the codimension-one separatrices between basins — precisely the fold and cusp loci that Thom's classification would later prove are the only generic ways such a separatrix can appear in a landscape with one or two control parameters. Waddington's pegs and guy-ropes turn out to be a physically-motivated sketch of a gene-regulatory network computing exactly the vector field whose integral curves the ball follows.

The same week that produced WP-30 also produced its correction pair, WP-31C and WP-31D, which examine a different switch — the AMPK–mTORC1–ULK1 autophagy decision — with the same landscape logic: is the observed bistability a real basin structure in a real potential, or a picture asserted without the underlying vector field being checked against a real biochemical network? Waddington's own discipline is the useful contrast. He drew the landscape confident that some real topography lay under it, and he was right; he did not additionally claim to know the topography's numerical slope, curvature, or eigenvalues, and on that point restraint served him better than a premature formula would have. The chapter on René Thom documents what happens when that restraint is dropped.

References