An embryo has no trouble with up and down. Gravity and the animal-vegetal axis handle that early. Front and back are established by the direction of gastrulation — cells moving, organizing, pulling the anterior from the posterior. These axes get anchored.
Left and right are different. Nothing in the physical environment reliably marks "left." There is no gravitational left, no chemical gradient drifting in from outside. For a period in development, the embryo is genuinely symmetric: everything on the left mirrors the right. And yet, in most vertebrates, the heart ends up on the left, the liver on the right, the gut coiled in a specific direction. The question is how that happens — how an organism with no external left-right reference ends up with one.
The answer is small enough to seem implausible.
In the embryonic node — a small pit at the midline of the embryo, present for a few hours during late gastrulation — cells grow cilia that rotate. Not wave, not beat in alternating arcs: rotate, like propellers, at around 600 rpm. Their structure (9+0, lacking the central pair of microtubules) is the same as what appears in immotile "primary" cilia found on most non-specialized cells. The assumption when these were first observed was that they couldn't be motile. They are.
The cilia are tilted about 40 degrees toward the posterior.
That tilt is the whole trick. A cilium rotating without any tilt produces a symmetric vortex — fluid spinning in a circle, no net flow anywhere. But a cilium tilted posteriorly rotates asymmetrically relative to the cell surface beneath it. When it sweeps leftward, the tip rises away from the surface; when it sweeps rightward, it moves close. Viscous drag near a stationary surface resists motion more than drag in open fluid. So the rightward sweep is retarded relative to the leftward swing. The result is a directed push to the left.
This generates nodal flow: a slow leftward current in the fluid filling the node, around 2–5 micrometers per second. And that flow carries something.
What it carries isn't a dissolved molecule drifting outward. It's membrane-wrapped packets — nodal vesicular parcels — containing Sonic hedgehog and retinoic acid, physically transported leftward by the current, then fragmented at the left edge of the node. The signal is delivered, not dispersed. From there, the cascade: Nodal protein (named for the structure, not the other way around) accumulates on the left side, activates Lefty (a feedback inhibitor), activates Pitx2. The left lateral plate mesoderm receives the signal. Organ development proceeds with left-right information now encoded in the molecular environment for the first time.
What happens when the cilia don't run?
Kartagener syndrome — also called primary ciliary dyskinesia — is a condition in which dynein arm defects leave cilia immotile throughout the body. The cilia are present and structurally intact; they just don't move. In the embryonic node, no rotation, no flow, no directed transport.
The result: roughly half of Kartagener patients have situs inversus totalis — complete mirror-image arrangement of all organs, heart on the right, liver on the left, gut coiled the other way. The other half have normal arrangement.
Not all reversed. Not even mostly reversed. Roughly half each way.
This distribution is not noise. It is the absence of signal. When the directional mechanism fails, the outcome that would have been resolved non-randomly is instead resolved by chance. The 50/50 split tells you, with unusual clarity, that nodal flow was the only systematic influence on laterality. There was no backup mechanism. No secondary axis. When the cilia stopped, the information stopped with them.
Situs inversus totalis, when complete, is largely compatible with healthy life. The mirror-image heart pumps; the mirror-image gut digests. Everything is reversed but coordinated — the organs still fit together. The problem comes with heterotaxy, sometimes called situs ambiguus, where the cascade is partially disrupted rather than absent. In that case, organs randomize independently. The heart might end up on the right while the gut coils rightward too; or the left lung develops three lobes (a right-side pattern) while the heart sits normally. The mismatches cause serious cardiac defects, because the left-right program coordinates multiple structures that need to be compatible. Complete reversal is one thing; inconsistent reversal is another.
The tilt of the cilia comes from planar cell polarity: the mechanism that aligns cell structures relative to the tissue's anteroposterior axis. Basal bodies — the anchoring structures from which cilia grow — are positioned toward the posterior of each node cell, and this positioning causes the resulting cilia to tilt posteriorly when they rotate. The A-P axis, already established, is the reference from which left-right is derived.
The direction of the rotation itself comes from the molecular handedness of the dynein motors. All biological amino acids are left-handed; the proteins built from them have consistent chirality. The rotation is right-handed because the molecular machinery is right-handed. This is universal and invariant — every vertebrate embryo's cilia rotate the same way because every vertebrate's proteins are chiral the same way.
So the chain goes: molecular chirality → dynein rotation direction → cilia tilt relative to A-P axis → net leftward flow → Nodal cascade → organ laterality. The embryo derives left-right from two things it already had: the direction the anterior was established, and the handedness of its own proteins. Neither of those is "left." Together, via the geometry of a tilted rotating cylinder near a wall, they construct it.
Before the cilia in the node start running, there is no left. Not in the sense that left is hidden or ambiguous — there is no fact of the matter yet about which side is which. The cilia establish the fact. The distinction doesn't pre-exist the mechanism that creates it.
The Kartagener result confirms this in the starkest possible way. The 50/50 split isn't telling you that the mechanism failed to detect a pre-existing left. It's telling you that the thing which would have made one side left failed to run, so neither side became left, and the downstream structures that needed the signal distributed themselves by chance. You can know which way a Kartagener patient's heart sits and learn nothing about the rest of their organs. Not because the information is encrypted. Because it was never produced.
What I find difficult to hold onto is the temporal structure of it: a few hours, a small pit of cells, cilia rotating at 600 rpm in a thin film of fluid. Then it's over. The node resorbs. The cilia stop. The organism continues developing, and its left side is set, and nothing that comes after preserves any record of the mechanism that set it. The adult has a left. The adult doesn't have cilia in their embryonic node. The construction site is gone.