Saltatory Conduction
A squid giant axon, 500 micrometers wide, conducts an action potential at about 25 meters per second. A human peripheral motor axon is only 10 micrometers wide. Yet it conducts at up to 100 meters per second. That is four times faster in a fiber fifty times thinner. This is not a small engineering curiosity. It is the reason a person can react to a hot stove in milliseconds. It is also why a giraffe's leg muscles fire almost as fast as its neck muscles, despite the two-meter cabling difference. And it is why certain diseases that strip axons of their insulation cause catastrophic slowing of nerve conduction. That slowing can sometimes be fatal, rather than a modest deficit. The solution vertebrates evolved is Myelination and the Insulation of Axons: wrapping axons in fatty sheaths. This forces the electrical signal to travel differently than it does in bare, unmyelinated fibers. But insulation alone would simply prevent signaling. A wire wrapped in plastic still needs bare contact points to do useful electrical work. The genius of the myelinated axon is that the insulation is deliberately interrupted at regular intervals. These interruptions occur at the nodes of Ranvier. It is at these tiny bare patches, only about a micrometer long, that essentially all of the axon's excitable machinery is concentrated. The action potential does not creep smoothly down the membrane as it does in an unmyelinated fiber. Instead, it appears to leap from node to node. This process is named saltatory conduction, from the Latin saltare, to leap. This article explains why myelin alone is insufficient. It shows how the alternation of insulated internodes and excitable nodes works. This alternation converts continuous propagation into a much faster jumping process. It also shows how classical electrophysiology proved this actually happens, rather than merely being a plausible story. It works through the quantitative logic — cable theory, conduction velocity scaling, and the space and time constants. These factors make saltatory conduction roughly 10 to 100 times faster than conduction in an unmyelinated axon of the same diameter. It does this while using a small fraction of the metabolic energy and axonal volume. Understanding this mechanism sets up the next event in the signaling chain: what happens when the action potential finally reaches the axon terminal and must cross the synaptic cleft.
- Myelin sheaths insulate internodes; voltage-gated Na⁺ channels cluster almost exclusively at nodes of Ranvier, spaced roughly every 0.2–2 mm.
- Under myelin, current spreads passively (electrotonically) with little loss because myelin raises membrane resistance and lowers capacitance.
- The action potential is regenerated only at nodes, so it appears to "jump" node to node — saltatory conduction.
- Conduction velocity scales roughly linearly with fiber diameter in myelinated axons, versus only with the square root of diameter in unmyelinated axons — a much better speed return on added girth.
- Huxley and Stämpfli's 1949 experiments on isolated frog nodes directly recorded and mapped current flow, confirming node-to-node regeneration.
- Demyelinating diseases (multiple sclerosis, Guillain-Barré syndrome) slow or block conduction by disrupting this exact mechanism, which is why nerve conduction studies are diagnostic tools.
Myelin does not stop the electrical signal from spreading — it forces it to spread passively and efficiently between nodes, so the metabolically expensive process of regenerating the action potential happens only at the sparse nodes of Ranvier, making conduction dramatically faster and cheaper than continuous propagation.
The core idea
In an unmyelinated axon, the action potential propagates by continuous conduction: at every point along the membrane, depolarization opens voltage-gated Na⁺ channels, Na⁺ rushes in, and the resulting local circuit current depolarizes the immediately adjacent patch of membrane to threshold, which then does the same to the next patch, and so on. Every single micrometer of membrane must reach threshold and fire its own full action potential. This is reliable but slow, because each cycle of channel opening, ionic flux, and local depolarization takes real time, and it must repeat at every point, with the amplitude of the traveling signal never decaying because it is actively regenerated everywhere. Saltatory conduction exploits a different strategy: separate the job of insulating and passively conducting current from the job of actively regenerating the signal, and assign them to different physical structures. The myelin sheath, formed by oligodendrocytes in the central nervous system and Schwann cells in the periphery, wraps the axon in up to 100 or more concentric lipid bilayer wraps over stretches called internodes, typically 0.2 to 2 millimeters long depending on axon diameter and species. Myelin dramatically increases the electrical resistance across the membrane and dramatically decreases its capacitance, because the many wraps are electrically nearly equivalent to increasing membrane thickness. Between internodes lie the nodes of Ranvier, tiny gaps roughly 1 micrometer long where the axonal membrane is bare and where voltage-gated Na⁺ channels are packed at very high density — often more than 1,000 channels per square micrometer, compared to roughly 25 per square micrometer in unmyelinated axon membrane. Because the internodal membrane is a poor generator (few channels) but an excellent passive conductor (high resistance, low capacitance), current injected at one node spreads passively — electrotonically — down the axon core and across the internodal membrane with very little loss, arriving at the next node still strong enough to bring that node's dense Na⁺ channel population to threshold. The action potential is therefore only actively regenerated at nodes; between nodes it travels as a decremental but fast passive spread. Functionally the impulse appears to jump from node to node — hence "saltatory" — even though physically the depolarizing current is continuous, just partly active and partly passive.
Mechanism
Step 1 — Myelin converts the internode into a low-loss cable. The passive spread of voltage along any axon segment is described by cable theory, which treats the axon as a leaky electrical cable characterized by two key parameters: the length constant (λ, "lambda") and the time constant (τ, "tau"). The length constant is λ = √(rₘ / rᵢ) where rₘ is membrane resistance per unit length and rᵢ is internal (axoplasmic) resistance per unit length. A larger λ means voltage decays more slowly with distance. Myelin increases rₘ by a factor roughly proportional to the number of wraps (often 100- to 300-fold), which increases λ substantially without changing axon diameter or axoplasmic resistance at all. Myelin also decreases membrane capacitance Cₘ, and because the membrane time constant τ = rₘ·Cₘ, the two effects partially offset in τ, but the dominant benefit is that far less current leaks out across the internodal membrane, so much more of the injected charge survives to reach the next node.
Step 2 — Sodium channels cluster at nodes, driven by specific axon-glia adhesion. During development, myelinating glia signal to the axon to cluster voltage-gated Na⁺ channels (chiefly the Naᵥ1.6 isoform in mature nodes) at nascent nodal gaps, anchored by cytoskeletal scaffolding proteins (ankyrin-G, βIV-spectrin) and paranodal adhesion complexes (including Caspr, contactin, and neurofascin-155) that seal the paranodal junction and prevent channels from diffusing into the internode. This creates a sharply discontinuous distribution of excitability: nodes are hyperexcitable, internodes are nearly inexcitable, by design.
Step 3 — Passive spread depolarizes the next node to threshold. When a node fires an action potential, the local circuit current it generates flows down the axon core (axoplasm), across the resistive but far-from-perfect internodal membrane, and out at the next node. Even though this spread is passive and therefore decremental (it weakens with distance, unlike active propagation), the myelinated internode is such an efficient low-leakage cable that the signal arrives at the next node with 3- to 5-fold more amplitude than the threshold voltage needed to fire it — a substantial safety factor.
Step 4 — The next node regenerates a full-amplitude spike. Because the depolarization exceeds threshold at the node, the dense population of nodal Na⁺ channels opens, generating a new, full-sized action potential exactly as if starting fresh. This regenerated spike then repeats the same passive spread across the next internode. The action potential is thus a series of discrete, full-amplitude "restarts" at nodes, linked by fast passive electrotonic jumps across internodes — the alternation is what makes conduction look saltatory rather than continuous.
Step 5 — Speed emerges from skipping most of the axon's length. Because active regeneration (the slow, channel-kinetics-limited part) only needs to happen at nodes spaced roughly 100 times the axon diameter apart, and passive spread (the fast part, limited mainly by cable properties and essentially instantaneous compared to channel gating) covers the rest, the net conduction velocity is far higher than continuous conduction over the same distance. Quantitatively, conduction velocity in myelinated fibers scales approximately linearly with axon diameter (v ≈ 6 × diameter in micrometers, in meters per second, for typical mammalian fibers), whereas in unmyelinated fibers velocity scales only with the square root of diameter (v ≈ 1.7 × √diameter). This different scaling law is the quantitative heart of why myelination is such an efficient evolutionary solution to the problem of fast, thin axons: doubling diameter roughly doubles speed in myelinated fibers but only increases it by about 40% in unmyelinated ones.
Why safety factor matters. The ratio of the depolarization actually arriving at a node to the threshold voltage needed to fire it is called the safety factor, typically 3 to 5 in healthy myelinated fibers. This buffer is what allows conduction to survive some degree of internodal leak or a slightly weakened signal without failing — but it also explains why demyelinating disease is so dangerous: as myelin thins or is stripped away, rₘ falls, λ shrinks, and less current reaches the next node. Once the arriving depolarization drops below threshold, conduction block occurs entirely, not just slowing — the axon simply stops transmitting past that point, even though the axon itself is intact.
The evidence
The idea that conduction "jumps" between nodes was proposed on largely indirect grounds in the 1920s and 1930s by Ralph Lillie and later refined theoretically by Ichiji Tasaki, who in the late 1930s used electrical recordings from frog myelinated nerve fibers to argue that the region under myelin was electrically inert and that current entered and left the axon predominantly at the nodes. But the decisive, direct demonstration came from Andrew Huxley and Robert Stämpfli in 1949, working on single isolated myelinated nerve fibers from frogs (Xenopus and Rana).
What they did. Huxley and Stämpfli dissected single myelinated axons from frog sciatic nerve and mounted them in a specialized chamber that allowed them to isolate short segments of the fiber — including individual internodes and individual nodes — electrically from one another using oil gaps or sucrose gaps, a technique that lets an experimenter apply current or record voltage from a defined stretch of axon in isolation. This let them directly measure the local circuit currents flowing at and between specific, identified nodes of Ranvier, rather than only observing the compound action potential of the whole nerve trunk from the outside.
What they observed. Three findings together nailed the mechanism down: 1. When they recorded the action potential's local currents along the fiber, current flowing into the axon (inward, depolarizing) was concentrated tightly at the nodes, while over internodal stretches, current flow was almost entirely of the outward, passive "leak" kind expected from a well-insulated cable, not the large regenerative inward current expected if Na⁺ channels were active there. 2. They could measure conduction time between two identified adjacent nodes and show that essentially all of the delay in propagation was incurred at nodes (where the relatively slow kinetics of channel opening and the action potential's rise time occur), while the time to traverse the internode was extremely short and consistent with passive, nearly instantaneous electrotonic spread. 3. When they experimentally blocked or altered the properties of a single node (for instance, by locally changing the ionic environment), conduction across that node could be selectively delayed or blocked, while adjacent internodes remained unaffected — showing the excitable machinery really is segregated to nodes, not smeared continuously along the fiber.
What it proved. This directly confirmed that the axon does not conduct by continuously regenerating the spike at every point, but rather by regenerating it only at discrete nodal loci separated by internodes that behave as passive cables. It converted saltatory conduction from an inference based on anatomy (myelin exists, nodes exist, therefore maybe...) into a measured electrophysiological fact: real current maps showing regeneration localized to nodes and passive spread in between. Huxley and Stämpfli's data also allowed the first direct experimental estimate of internodal conduction time versus nodal delay, giving quantitative support to the cable-theory account of why myelinated fibers conduct so much faster. A complementary and equally important body of evidence came later from voltage-clamp and immunohistochemical studies (from the 1970s onward, notably by Stephen Waxman, Peter Shrager, Matthew Rasband and others) directly visualizing the molecular distribution of Na⁺ channels using antibodies and electron microscopy, confirming the channel-density asymmetry (roughly 1,000–2,000/µm² at nodes versus roughly 20–25/µm² in unmyelinated axon membrane, and even lower under intact myelin) that the electrophysiology had implied. Freeze-fracture electron micrographs from the 1970s (Rosenbluth and others) directly showed clusters of intramembranous particles — visible protein clusters corresponding to channel proteins — concentrated exactly at nodal axolemma, giving structural confirmation of the physiological measurements.
Worked examples
Example 1 — Comparing conduction velocity across fiber types (a clinical nerve-conduction calculation). A clinical neurophysiologist is evaluating two peripheral nerve fibers in a nerve conduction study. Fiber A: myelinated, diameter 12 µm. Fiber B: unmyelinated, diameter 1.0 µm (representative of a C-fiber carrying dull pain/temperature). Using the standard approximations:
- Myelinated: v (m/s) ≈ 6 × diameter (µm)
- Unmyelinated: v (m/s) ≈ 1.7 × √diameter (µm)
Fiber A: v ≈ 6 × 12 = 72 m/s (consistent with a large myelinated Aα/Aβ motor or proprioceptive fiber). Fiber B: v ≈ 1.7 × √1.0 = 1.7 m/s (consistent with C-fiber pain conduction). Ratio of velocities: 72 / 1.7 ≈ 42-fold faster, despite Fiber A being only 12-fold larger in diameter. This illustrates the nonlinear payoff of myelination directly: not only is the myelinated fiber faster in absolute terms, it is faster per unit of diameter invested, because it obeys a linear scaling law rather than a square-root one. Clinically, this is exactly why touching something painfully hot produces a fast withdrawal reflex (mediated by fast Aβ/Aδ myelinated afferents and motor efferents) followed noticeably later by the slow, burning, prolonged pain sensation (carried by unmyelinated C-fibers) — the "double pain" phenomenon taught in somatosensory physiology is a direct behavioral readout of this velocity gap.
Example 2 — Diagnosing conduction block from safety factor loss (a clinical reasoning case). A 34-year-old presents with rapidly ascending weakness over several days, areflexia, and normal sensation, following a gastrointestinal infection — a classic presentation of Guillain-Barré syndrome, an autoimmune demyelinating polyneuropathy in which antibodies attack peripheral myelin (and sometimes nodal/paranodal proteins directly). Nerve conduction studies show:
- Distal motor latency prolonged.
- Conduction velocity reduced from a normal ~55 m/s to 20 m/s in some segments.
- In some nerve segments, stimulating proximal to a lesion produces almost no evoked muscle response at all, despite a normal response when stimulating distal to the lesion (a finding called "conduction block").
Reasoning through the mechanism: demyelination increases internodal capacitance and decreases internodal resistance rₘ (myelin wraps are lost), which shrinks the length constant λ = √(rₘ/rᵢ). A smaller λ means the passive depolarization arriving at the next node decays more steeply with distance. Where demyelination is partial, the arriving signal is weaker but still exceeds the node's threshold — conduction continues but slower (explaining the reduced velocity and prolonged latency). Where demyelination is severe enough that the arriving depolarization falls below threshold, the axon fails to regenerate a spike at that node at all — an all-or-none conduction block, even though the axon itself may be structurally intact and capable of recovering once remyelination (partly carried out by surviving or newly proliferated Schwann cells) restores rₘ. This is why Guillain-Barré syndrome, despite sparing the axon itself in its typical demyelinating form, can cause profound, even life-threatening, motor and respiratory weakness — the message home is present, but the electrical relay system that carries it has failed.
Common misconceptions & traps
| Misconception | Why it's wrong |
|---|---|
| "The action potential physically skips over the myelinated internode with no current flow at all." | Current does flow continuously through and across the internode — it just flows passively (electrotonically), not via active channel-mediated regeneration. Nothing is literally skipped; the regenerative event is what is confined to nodes. |
| "Myelin makes the axon a perfect insulator, so no current is lost under the sheath." | Myelin greatly reduces but does not eliminate current loss across the internodal membrane; it raises resistance and lowers capacitance, improving but not perfecting the cable. Some decrement is normal and is compensated by the nodal safety factor. |
| "Thicker myelin alone, without proper node spacing, guarantees fast conduction." | Conduction velocity depends on the interaction between internode length and axon diameter (both scale together during development); nodes spaced too far apart relative to diameter would let the signal decay below threshold before reaching the next node. |
| "All vertebrate axons are myelinated, and myelination is the only way to speed up conduction." | Many vertebrate axons (e.g., unmyelinated C-fibers) remain unmyelinated; invertebrates like squid achieve speed instead by growing enormous axon diameter, since axoplasmic resistance falls with diameter — a different solution to the same problem. |
| "Nodes of Ranvier are just gaps — passive weak points in the axon." | Nodes are metabolically and molecularly the most specialized, channel-dense, energetically active regions of the entire axon; they are the opposite of a weak point, they are the powerhouse of regeneration. |
The scientist behind it
Sir Andrew Fielding Huxley (1917–2012) is best known jointly with Alan Hodgkin for the 1952 Hodgkin-Huxley equations describing the ionic basis of the action potential, work that won them the 1963 Nobel Prize in Physiology or Medicine (shared with John Eccles). But three years before that landmark paper, in 1949, Huxley collaborated with the Swiss physiologist Robert Stämpfli on a less famous but foundational study: directly recording local circuit currents at individual, identified nodes of Ranvier in isolated frog myelinated fibers. Huxley's characteristic combination of rigorous quantitative electrophysiology and careful, almost artisanal single-fiber dissection technique — the same meticulousness that let him and Hodgkin later use the squid giant axon to nail down ionic conductance mechanisms — was exactly what made the node-by-node current mapping possible. Stämpfli, working in Bern, had developed some of the key isolated-fiber and local-perfusion techniques that made recording from single, tiny internodes and nodes technically feasible at all, a nontrivial experimental feat given the micrometer scale of a frog node.
Origins · 1 minAndrew Fielding Huxley and how this was discovered
Trained initially in physiology at Cambridge, Huxley built custom voltage-clamp apparatus by hand and treated the axon as a physical, quantifiable electrical system rather than a black box, a philosophy that unified his 1949 nodal current work with the 1952 Hodgkin-Huxley model of the action potential.
Where this matters today
Diagnosing and monitoring demyelinating disease. Nerve conduction studies and electromyography, routine clinical tools, directly measure conduction velocity and detect conduction block; a slowed or blocked signal is the electrophysiological fingerprint of demyelination, used to diagnose and stage multiple sclerosis, Guillain-Barré syndrome, chronic inflammatory demyelinating polyneuropathy (CIDP), and hereditary neuropathies like Charcot-Marie-Tooth disease.
Drug and therapy design targeting nodes and channels. 4-aminopyridine (dalfampridine), a K⁺ channel blocker, is FDA-approved to improve walking speed in multiple sclerosis patients partly by prolonging the depolarization at demyelinated, current-leaky segments, effectively boosting the safety factor enough to restore some conduction across partially demyelinated stretches.
Understanding local anesthetic and toxin action. Because nodes concentrate Na⁺ channels, local anesthetics (lidocaine, procaine) and biological toxins (tetrodotoxin, saxitoxin) that block voltage-gated Na⁺ channels exert an outsized effect at nodes; clinically, a local anesthetic needs to block only a few consecutive nodes (three is typically sufficient) to stop conduction entirely along a myelinated fiber, since regeneration cannot occur at intervening unblocked nodes if the neighboring blocked stretch prevents the depolarization from arriving above threshold.
Guiding remyelination research and regenerative medicine. Efforts to promote remyelination after spinal cord injury or in progressive multiple sclerosis (using oligodendrocyte precursor cell transplantation or pharmacological pro-differentiation agents) are explicitly aimed at restoring the nodal/internodal architecture described here, because restoring myelin without correctly reforming clustered nodal Na⁺ channel domains would not restore functional saltatory conduction.
Comparative and evolutionary neuroscience. Understanding why vertebrates evolved myelin (letting brains and bodies scale up without needing squid-sized axons) versus why invertebrates like squid and earthworms instead evolved giant-diameter axons for speed illuminates broader principles of convergent evolution under the shared physical constraint of cable theory — an elegant case where the same problem (fast axonal signaling) has two very different physical solutions, both explainable from the identical mathematics.
Ask the tutor: How would conduction velocity and safety factor change if a single internode were selectively demyelinated while all its neighbors stayed healthy?
Why doesn't the action potential just decay away entirely as it passively crosses the internode?
Because myelin makes the internodal membrane behave as a very low-leakage cable: it raises transmembrane resistance (fewer ion channels, thick insulating wraps) and lowers capacitance, so the length constant λ becomes large — often several millimeters, versus roughly 0.1–1 mm in bare unmyelinated membrane. Passive spread always decays with distance according to cable theory, but because λ is so large relative to typical internode lengths (0.2–2 mm), only a modest fraction of the signal is lost before reaching the next node. The signal arriving there is still 3- to 5-fold above threshold (the safety factor), which is why it can reliably fire a brand-new, full-amplitude action potential rather than fading into nothing.
Do all vertebrate neurons use saltatory conduction?
No. Many axons, especially small-diameter ones carrying slow, non-urgent information — such as C-fibers conveying dull, aching pain, temperature, and some autonomic signals — remain unmyelinated their entire length and conduct continuously and slowly (roughly 0.5–2 m/s). Myelination is metabolically and developmentally costly, requiring glial wrapping and precise molecular organization of nodes, so evolution appears to reserve it for axons where speed is functionally critical: motor neurons, proprioceptive afferents, and fast-reacting sensory and reflex pathways, while slower, low-priority information travels unmyelinated.
How do voltage-gated Na⁺ channels end up clustered so precisely at nodes and nowhere else?
Clustering is driven by molecular signaling between the axon and the myelinating glial cell, not by the axon alone. As Schwann cells or oligodendrocytes wrap and mature, they organize paranodal junctions using adhesion proteins (Caspr and contactin on the axon, neurofascin-155 on the glial loop) that physically seal off the region flanking the node, acting like fences. Meanwhile, cytoskeletal scaffold proteins (ankyrin-G, βIV-spectrin) anchor Na⁺ channels specifically within the nodal gap itself. This is an active, developmentally regulated process — disrupting any of these adhesion or scaffold proteins experimentally causes Na⁺ channels to become diffusely distributed rather than clustered, and conduction velocity drops correspondingly, directly demonstrating the causal link between molecular clustering and saltatory speed.
Is saltatory conduction unique to vertebrates?
Essentially yes, among well-studied groups — true compact myelin with nodes of Ranvier is a vertebrate innovation, though some invertebrates (certain shrimp, copepods, and annelids) have independently evolved superficially myelin-like wrappings around axons that also increase conduction velocity, a striking case of convergent evolution driven by the same underlying cable-theory physics. Most invertebrates, however, achieve conduction speed by an entirely different route: growing axons to very large diameter, since axoplasmic resistance falls as diameter increases, which is why the squid giant axon (used by Hodgkin and Huxley for their foundational action potential work) can be roughly 500 µm wide.
What would happen if nodes of Ranvier were spaced too far apart?
If internodes were too long relative to axon diameter and myelin quality, the passively-spread depolarization would decay below threshold before reaching the next node, and conduction would fail — an internal conduction block, with no disease required. This is why internode length and axon diameter are developmentally coordinated (both scale up together as axons mature and myelinate more heavily), keeping the safety factor comfortably above 1 across the whole fiber. It also explains why demyelinated axons, which effectively have "internodes" stretched to whatever length remains unwrapped, are so prone to exactly this kind of failure.
How does saltatory conduction relate to what happens next at the synapse?
Saltatory conduction gets the action potential to the axon terminal quickly and reliably, but it does not by itself transmit information to the next cell — that handoff is a separate, chemically mediated process. When the regenerated spike reaches the final node before the terminal, it depolarizes the presynaptic membrane, opening voltage-gated Ca²⁺ channels and triggering neurotransmitter release, the subject of synaptic transmission. Saltatory conduction is thus the fast, reliable "delivery" mechanism; synaptic transmission is the separate "translation" step that converts the electrical signal into a chemical message for the postsynaptic cell.
Why is conduction velocity often quoted as roughly linear in diameter for myelinated fibers but only as the square root of diameter for unmyelinated fibers?
In unmyelinated fibers, the entire membrane must be actively regenerated point by point, and conduction velocity is limited mainly by axoplasmic resistance, which falls only with the square root of the cross-sectional area (hence diameter) — giving the v ∝ √d relationship. In myelinated fibers, both internode length and axon diameter increase together during development (thicker axons get proportionally longer internodes), and because passive spread across a well-insulated internode is comparatively fast, the rate-limiting step becomes more directly proportional to how far each "jump" can cover per diameter increment, empirically yielding an approximately linear v ∝ d relationship. This different scaling law is precisely why myelination gives a much better speed return per unit of axon girth invested.