Section 43–3 The drift speed

S-molecules / Drift velocity / Mobility

 

In this section, Feynman introduces the concepts of S-molecules, drift velocity, and mobility (electrical and generalized). Although the section title is “The drift speed,” Feynman later shifts to using “drift velocity” in the text. This transition aligns with the standard physics term because the phenomenon is inherently directional, arising from an asymmetric external force. Conversely, "drift speed" is simply the scalar magnitude of that vector, a term best reserved for contexts where directional orientation is irrelevant. 

 

1. S-molecules

“We shall refer to the “majority” molecules as the “background” molecules, and we shall call the molecules which are different from the background molecules “special” molecules or, for short, the S-molecules (Feynman et al, 1963).”

 

The term S‑molecules (special molecules) serves as a shorthand for identifying the particle under study, distinguishing it from background molecules without lengthy descriptions. This abstraction unifies the treatment of diffusion of gases, centrifugal separation, and ionic conduction, under a framework, providing an intuitive explanation of drift velocity. However, the term is nonstandard in the broader literature, where expressions such as test particles or tagged particles could be used. Its use may confuse readers and inadvertently suggest that S‑molecules possess unique properties, but it is merely an analytical description. Furthermore, the model assumes a dilute system in which the S-molecules neither disturb the background molecules nor interact with one another, which is unlikely to happen in the real world. Here’s a fun twist, in the Audio Recordings of this lecture, Feynman did not use the term “S‑molecules”, but relied instead on phrases like “this thing,” “heavy object,” or simply “it.” He did, however, explicitly introduce the “fresh start” assumption—the idea that molecules do not remember their acceleration after each collision.

 

The limitations of the “fresh start” assumption can be illustrated with a bowling ball analogy: a heavy particle moving through a sea of lighter molecules behaves like a bowling ball rolling through bowling pins—each impact alters its motion only slightly. The assumption of a "fresh start" after every collision is unrealistic because the particle’s momentum remains highly correlated from one collision to the next; a single collision is usually insufficient to erase the memory of its prior motion. Instead of adopting t as the mean time between consecutive collisions, we can reframe it as the momentum relaxation time—the characteristic time over which cumulative collisions dissipate the particle’s initial directed momentum. This approach preserves the structure of the simple formula while incorporating the cumulative effects of many collisions, making the model both intuitive and more realistic. A rigorous treatment of this process is provided by the Boltzmann equation, Langevin equation, or Green-Kubo relations.

 

“We shall not work out the details, but just state that the result is equivalent to replacing τ, the average collision time, by a new—and longer—τ which corresponds to the average ‘forgetting time,’ i.e., the average time to forget its forward momentum (Feynman et al., 1963).”

 

It is worthwhile to clarify the distinction between the mean collision time and the momentum relaxation time (Feynman’s "forgetting time"). The mean collision time is a kinematic quantity that measures the average time between consecutive collisions and characterizes the frequency of collision events. By contrast, the momentum relaxation time is a transport quantity that measures the average time required for cumulative collisions to “erase” a particle’s directed momentum and establish a new equilibrium. The two timescales are identical if each collision completely randomizes the particle’s momentum. In real gases, liquids, and plasmas, momentum is lost gradually through a succession of small-angle deflections. Consequently, the momentum relaxation time is typically longer than the mean collision time, serving as the appropriate parameter for describing drift velocity and mobility.

 

2. Drift velocity

“We say that there is a drift, superposed on its random motion. We would like to know what the speed of its drift is—its drift velocity—due to the force F (Feynman et al, 1963).”

 

A comprehensive definition of Drift velocity

Drift velocity is the average, directed velocity acquired by charged particles (such as electrons or ions) when subjected to an external electric field—a slow, systematic drift superimposed upon their much larger, random thermal motion (typically averages to zero in the absence of a field). Microscopically, it arises from the acceleration by the applied force (typically F= qE) during the brief intervals between collisions and molecular force from each collision. Macroscopically, the drift velocity is directly proportional to the applied electric field and is related to the electric current density by J = nqv, where n is the charge carrier density. Although the random thermal speeds of particles are typically orders of magnitude larger than the drift velocity, it is this tiny net motion that is responsible for macroscopic electrical conduction, diffusion, and other transport phenomena. In electrical circuits, this allows the drift velocity to bridge the microscopic molecular collisions to the measurable electrical currents in response to applied fields.

 

“So between the two collisions it has, on the average, a velocity one-half of the final velocity, so the mean drift velocity is ½ Fτ/m. (Wrong!) This result is wrong and the result in Eq. (43.13) is right, although the arguments may sound equally satisfactory…… Short times occur more often but make less contribution to the drift velocity because they have less chance “to really get going.” If one takes proper account of the distribution of free times between collisions, one can show that there should not be the factor ½ that was obtained from the second argument. The error was made in trying to relate by a simple argument the average final velocity to the average velocity itself (Feynman et al., 1963).”

 

Feynman warns that the drift velocity cannot be obtained by simply halving the final velocity, but he only mentions the “distribution of free times” without showing the calculation. A more instructive approach uses the exponential distribution of free-flight times. Although the mean time between collisions is the constant t, individual free-flight times t are random and vary from one collision to another. They follow the probability density function p(t) = (1/t)e^{-t/t}, which arises from the assumption that collisions occur randomly and independently—a Poisson process (Feller, 1968). Immediately after a collision, the probability that the next collision occurs between t and t + dt is p(t)dt. During this interval, the particle accelerates under a constant force F, gaining an additional velocity (F/m)t. Averaging over all possible free-flight times gives the drift velocity:

Thus, the drift velocity is determined by the mean of the exponential free-flight distribution—not by assuming that every molecule accelerates for the same duration. However, while this approximation is valid for dilute gases, it becomes less reliable in dense fluids or systems with correlated collisions.

 

Note: In the Audio Recordings (26 min: 55 sec) of this lecture, Feynman mentions that the ½ factor error could be due to the textbook of Resnick and Halliday, or others.

 

Alternatively, Feynman could have clarified that a molecule’s initial velocity immediately after a collision is not zero. More precisely, after each collision, a molecule moves with a random (initial) velocity v₀, which is typically several orders of magnitude larger than the drift velocity. Between successive collisions, the external force F imparts a small systematic velocity increment Dv = (F/m)t where t is the time elapsed since the last collision. The molecule’s actual (instantaneous) velocity is therefore: v = v₀ + (F/m)t. The crucial assumption is that the post-collision initial velocities are isotropically distributed (equally likely in any direction), meaning their statistical average is zero (<v₀> = 0). This does not imply that each molecule has zero initial velocity after a collision; rather, the random directions of many molecules cancel statistically. Thus, the drift velocity arises solely from the small, nonzero average produced by the force-induced term. The macroscopic drift represents a small directional bias superimposed on the much larger random thermal motion.

 

3. Mobility

“If in an electrical problem the force is written as the charge times the electric field, F=qE, then the constant of proportionality between the velocity and the electric field E is usually called the “mobility.” In spite of the possibility of some confusion, we shall use the term mobility for the ratio of the drift velocity to the force for any force. We write v_drift = μF (43.14). in general, and we shall call μ the mobility (Feynman et al., 1963).”

 

Perhaps Feynman could have distinguished between electrical mobility and generalized mobility. His discussion begins with an electrical problem in which the force on a charged particle is electrical (F = qE). In this context, the proportionality constant between the drift velocity and the electric field is the electrical mobility, typically denoted as m_e, giving the equation v_d = m_eE. Electrical mobility is a physical property that quantifies how fast a charged particle drifts through a medium under an applied electric field. Mathematically, it is defined as the ratio of the drift velocity to the electric field strength: m_e = v_d/E. Experimentally, it can be measured by several techniques, including electrical mobility analyzers (EMAs) or Hall-effect measurements. Electrical mobility is a key transport parameter in semiconductor and ionized gases, and it plays a central role in the design of electronic devices such as transistors, diodes, and solar cells.

 

Note: In his Lectures on Computation, Feynman (1996) explains: “Drawing on engineering practice, we can write the drift velocity v_d in terms of the mobility m of the charge carriers as v_d = m_eE, E is the electric field across the drain/source (p. 225).”

 

When Feynman introduces the general equation v_d = mF, the quantity m does not represent electrical mobility but a generalized (or mechanical) mobility. This transport coefficient relates the drift velocity to any applied force—electrical, gravitational, or chemical. Under an electric field, where F = qE, this equation becomes v_d = mF = m(qE). Comparing this with the definition of electrical mobility, v_d = μ_eE, gives: μ_e = qm. Thus, a particle’s electrical mobility is simply its generalized mobility multiplied by its charge. The generalized mobility is also the quantity that appears in the Einstein diffusion relation, D = mkT, where D is the diffusion coefficient. Expressed in terms of the electrical mobility, the relation becomes D = μ_ekT/q. In summary, the generalized mobility is the more general charge-independent transport coefficient, while the electrical mobility is a special case that describes a particle’s response specifically to an electric field.

 

Key Takeaways (In Feynman’s Tone):

1. The "Special Molecule" Idea

Suppose you've a gas with an enormous number of molecules, all flying about, colliding with one another, and changing their directions every instant. Trying to follow every molecule would be hopeless. Instead, we single out one molecule—the special molecule, or S-molecule—and ask only what happens to it. Maybe it’s heavier than the others, maybe it carries a charge, or maybe it’s being pushed by an electric force while the background molecules simply collide with it. By focusing on this one molecule, we can understand the basic ideas behind many transport phenomena—diffusion, electrical conduction, and sedimentation. It's not the whole truth—it's an approach to get started.

 

2. Where the Drift Comes From

Now consider what happens to the S-molecule. Between collisions it is accelerated by an applied force, but every collision sends it off again in a new random direction. The collisions randomize the motion; the external force biases the direction.

The result is a tiny average velocity—the drift velocity—superimposed on the molecule’s much larger random thermal motion. You can barely see it, but it's this small average motion that produces measurable phenomena such as electric current, diffusion, and sedimentation.

 

3. Mobility: How Easily a Molecule Drifts

How fast does the molecule drift for a given force? That depends on the mobility, which is how readily the molecule moves through a medium. Here's the catch: you have to be careful what you mean by "mobility."

• If you define it as the drift velocity per unit force, we'll call it the generalized mobility—it works for any force, electric or others: m = v_d /F .
• If you define it as drift velocity per unit electric field, we call that the electrical mobility: m_e = v_d /E.

They're simply related: μ_e = qm. Thus, the electrical mobility is just the generalized mobility multiplied by the charge. So if you know one, you know the other. It's not two different physical properties—it's the same mechanism, just calibrated for different kinds of forces.

 

The Moral of the Lesson: From Drift Velocity to Cholesterol Transport

Cholesterol does not circulate through the bloodstream as isolated molecules. Because it is a hydrophobic lipid, it must be packaged into spherical vehicles known as lipoproteins—such as low-density lipoproteins (LDL) and high-density lipoproteins (HDL). From a physicist’s perspective, these lipoproteins can be viewed as Feynman’s "special molecules" (S-molecules) moving through a complex fluid environment crowded with background molecules. Just like the S-molecule, each lipoprotein undergoes continual collisions with surrounding water molecules, plasma proteins, and blood cells, while external forces—whether drag, gravity or electrical—can affect its motion.

 

One laboratory technique that illustrates this idea is electrophoresis, in which an electric field causes charged molecules to migrate through a supporting medium. Because different types of lipoproteins possess distinct surface charges arising from their apolipoproteins, they experience different electrical forces. After a brief period of acceleration, the electrical force is balanced by the viscous drag of the medium, causing each molecule to reach a constant drift velocity. The ratio of this drift velocity to the applied electric field is called the electrophoretic mobility. Although electrophoresis is used mainly for specialized lipoprotein analysis rather than routine cholesterol testing, it provides a real-world example of Feynman’s discussion of drift velocity and mobility.

 

Note on Density: The terms “low‑density” and “high‑density” originated from the separation of these molecules via an analytical ultracentrifugation. In a centrifuge, lipoproteins are segregated by their density—not by electrophoretic mobility.

 

The "Drift Velocity" Analogy for Atherosclerosis

The concept of drift velocity also provides a useful analogy for understanding atherosclerosis. LDL molecules do not simply “flow” passively through the bloodstream; rather, they undergo continual random motion while being transported by the flowing blood. The probability that a LDL molecule interacts with the arterial wall increases with its concentration and the condition of the vessel wall.  Consider a mechanical analogy: a healthy artery has a smooth endothelial lining (like a new road), while a damaged one has a rough, permeable surface where molecules can become trapped. Over time, as LDL molecules “drift” through the bloodstream, a higher concentration of them—akin to high traffic density—leads to more interactions with the artery wall. When too many LDL molecules accumulate in a damaged artery wall, they can become trapped and oxidized, triggering an inflammatory immune response that contributes to atherosclerosis. For this reason, many scientists consider the number of LDL molecules to be a better indicator of cardiovascular risk than the total amount of cholesterol—but LDL molecules can be further classified into several subclasses with differing biological properties.

 

The Science of Temporary LDL Increase After Intense Exercise

Intense exercise can cause a temporary increase in measured LDL cholesterol levels. This short term elevation is generally attributed to two physiological mechanisms rather than to any harmful effect of exercise. 

Hemoconcentration: Strenuous exercise, such as long-distance running, results in substantial fluid loss through sweating and respiration. The reduction in plasma volume temporarily increases the concentration of the blood, meaning that about the same number of LDL molecules remain within a smaller volume. Consequently, a lipid panel may show a transient increase in measured LDL concentration. In one study, total plasma cholesterol rose by 24% immediately post-exercise, with LDL-C increased by about 20% (Krum et al., 1991).

 

These changes are short-term physiological responses to intense exercise rather than evidence that exercise is harmful. As hydration is restored and the body recovers, lipid concentrations generally return to their usual baseline. More importantly, the long-term effects of regular physical activity maintain healthy lipid metabolism and reduce cardiovascular risk.

 

Summary and Takeaways

Distinguishing short-term transport phenomena from long-term equilibrium behavior is important in both physics and medicine. The short-term “rise” in LDL after intense exercise is a normal physiological response—not a sign that exercise is harming your heart. When you are well-rested and hydrated, regular exercise has a proven beneficial effect on cholesterol profiles, which is why doctors recommend it for heart health.

To ensure clinical accuracy and eliminate transient variables, cholesterol levels should be measured under consistent conditions—well-rested, fully hydrated, and following a standard 12-hour fast. Thus, doctors should advise the public to avoid strenuous or exhaustive exercise for 12-24 hours prior to a blood test. This precaution ensures that temporary transport dynamics do not obscure the patient’s true, steady-state lipid profile.

 

Review Questions:

1. Why did Feynman introduce the concept of the S-molecule, and what is the purpose of the "fresh start" assumption?

2. Why is it incorrect to assume that the average drift velocity is simply one-half of the final velocity between collisions?

3. How would you distinguish between generalized (mechanical) mobility and electrical mobility?

 

References

Feller, W. (1968). An Introduction to Probability Theory and Its Applications (Vol. 1, 3rd ed.). Wiley.

Feynman, R. P. (1996). Feynman lectures on computation. Reading, Massachusetts: Addison-Wesley.

Feynman, R. P., Leighton, R. B., & Sands, M. (1963). The Feynman Lectures on Physics, Vol I: Mainly mechanics, radiation, and heat. Reading, MA: Addison-Wesley.

Krum, H., Conway, E. L., & Howes, L. G. (1991). Acute effects of exercise on plasma lipids, noradrenaline levels and plasma volume. Clinical and experimental pharmacology and physiology, 18(10), 697-701.

Section 43–2 The mean free path

Mean free path / Collision cross section / l-σ relation  

 

In this section, Feynman discusses the mean free path, the collision cross section, and the l-σ relation that connects them. These ideas resolve a paradox in kinetic theory of gases: if gas molecules move at high speeds, why do gases mix and diffuse slowly? The answer is that molecules do not travel unimpeded in straight lines, but they undergo incessant collisions in a chaotic random walk. The concept of mean free path was introduced by Clausius in his 1858 paper On the mean lengths of the paths described by the separate molecules of gaseous bodies, where he provided the first quantitative description of the average distance traveled by a molecule between successive collisions.

 

1. Mean free path

“If we say that the average time between collisions is τ, and that the molecules have a mean velocity v, we can expect that the average distance between collisions, which we shall call l, is just the product of τ and v. This distance between collisions is usually called the mean free path:

Mean free path l = τv (Feynman et al., 1963).”

 

A Comprehensive Definition of Mean Free Path

The mean free path (l) is the average distance a particle travels between successive collisions in a dilute gas. It is a statistical quantity that links the microscopic collisions to macroscopic transport phenomena such as diffusion, viscosity, and thermal conductivity.

Kinematic Perspective: It is the product of the particle's mean speed (v) and the mean collision time (t), l(v) = v t(v), which is directly related to the Maxwellian speeds. 

Geometric Perspective: For identical hard-spheres with diameter (d) and number density (n), the collision cross section is s = pd^2. Under the stationary target assumption, the mean free path is l =1/ns—it is inversely proportional to both number density and collision cross-section.

Statistical Perspective: The collision-free distances follow an exponential distribution, P(x)=(1/l)e^{-x/l], where P(x) is the probability that a particle travels a distance x. Thus, most particles travel less than l, while relatively fewer travel longer without any collision.

Thermodynamic Perspective: the mean free path also depends on the macroscopic states, e.g., at constant pressure, l = kT/(Ö2)sp, so increasing the temperature increases the mean free path.

       Thus, the mean free path is more than a simple average distance between collisions. It is a unifying parameter that connects the kinematics of particle motion, the geometry of molecular encounters, the statistics of random collisions, and the thermodynamic state of the gas.

 

The Stationary-Target model

Feynman’s formula is based on a simplifying assumption: all scattered molecules (targets) are treated as if they were stationary relative to the moving molecule. In reality, every molecule is moving with a high speed. What determines the collision rate is therefore not the speed of a single molecule relative to the laboratory, but the relative speed between pairs of molecules.

Suppose two identical molecules have velocity vectors v₁ and v₂, the relative velocity is

v_rel = v₁ - v₂

The mean-square relative speed is

<v_rel²> = <(v₁ - v₂)²>

Expanding the squares gives,

<v_rel²> = <v₁²> + <v₂²> - 2<v_1·v₂>

In thermal equilibrium, the velocities of different molecules are uncorrelated, so <v_1·v₂> = 0
Since the molecules are identical,

<v₁²> = <v₂²> = <v²> Þ  <v_rel²> = 2<v²>.

Taking the square root gives the mean relative speed

v_rel = (Ö2)v

where (v) denotes the corresponding root-mean-square molecular speed.

 

Physically, this means that collisions occur Ö2 times more frequently than predicted by the stationary-target model because every moving molecule contributes to the relative motion.

 

Note: In the audio recording (15 min: 45 sec) of this lecture, Feynman says something like: “Let’s say the nuclides are standing still…”, but this assumption is not included by the editors. The stationary-target model is intended as a simplification rather than an accurate derivation.

 

2. Collision cross section

“By ‘collision cross section’ we mean the area within which the center of our particle must be located if it is to collide with a particular molecule. (Feynman et al., 1963).”

The collision cross section, usually denoted by s, is a measure of the probability that two particles will collide or interact when they approach one another. Although it has the dimensions of an area, it should not always be interpreted as the particle's actual physical size. In the classical hard-sphere model, the collision cross-section is a fixed geometric area, which is independent of direction of approach. However, in Maxwell’s (1867) kinetic theory, the force between molecules is assumed to vary inversely as the fifth power of their separation. Under this assumption, the collision cross-section is no longer constant but varies inversely with the relative speed of the molecules. In general, the collision cross‑section is not a single, immutable number but a dynamical quantity—an effective interaction area that depends on the forces, speeds, and nature of the particles involved. It is the fundamental parameter that governs how frequently particles scatter, how far they travel between collisions, and ultimately how gases conduct heat, diffuse, and flow.

“If molecules were little spheres (a classical picture) we would expect that σ_c=π(r₁+r₂)², where r₁ and r₂ are the radii of the two colliding objects (Feynman et al., 1963).”

A Feynman-Style Explanation of the Collision Cross Section

Imagine a large male dancer standing in the middle of a crowded dance floor. Let us say he has a radius R. Moving through the crowd is a smaller female dancer with radius r. Both are free to move, but to keep things simple, let us freeze the large dancer and let only the smaller one move.

Now the question is: When do they “collide”?

 

A casual observer might say, "When her elbow touches his belly." But for a physicist, that’s sloppy thinking. We can ignore the complicated shapes of bodies and keep track of their centers. At the instant the two dancers just graze one another—when her shoulder just brushes his arm—the distance between their centers is neither R nor r. It is the sum: R + r. Why? Her center is one r away from her edge, and his center is one R away from his edge. When the two surfaces just touch, the centers must be separated by the sum of their radii.

 

Now comes the clever trick.

Keeping track of two moving objects with finite size is inconvenient. So we cheat. We shrink the smaller dancer down to a mathematical point. Of course, doing that would change the physics—however, we can transfer the smaller dancer’s radius to the larger dancer.

Mathematically, the larger dancer acquires an effective radius R+r.

 

In other words, we can idealize the female dancer as a point zipping around, and the male dancer becomes am inflated stationary target. She will collide with him if her center ever reaches a circle of radius R+r centered on him.

 

What's the collision cross-section then? 

It is simply the area of that inflated target—the area of a circle with radius (R+r):

s = p(R+r)²

What does it mean?

The collision cross section is determined not by the size of either particle alone, but by the combined radius of both particles. Collisions depend on how close the centers of the particles can approach, not simply on the so-called physical size of one particle.

 

However, this is only the beginning. One may modify the explanation or analogy to include, for example, the interaction forces, relative speeds, and quantum effects.

 

3. l-σ relation 

“… if we write it as σ_cn₀l=1. (43.12). This formula can be thought of as saying that there should be one collision, on the average, when the particle goes through a distance l in which the scattering molecules could just cover the total area. In a cylindrical volume of length l and a base of unit area, there are n₀l scatterers; if each one has an area σ_c the total area covered is n₀lσ_c, which is just one unit of area (Feynman et al., 1963).”

 

According to Feynman, the formula σnl = 1 can be visualized as follows: a particle traveling a distance l sweeps out a cylindrical volume of cross-sectional area σ. On the average, this cylinder contains exactly one scattering target, so one collision is expected over that distance. However, this formula rests on a hidden assumption: the scattering molecules are treated as stationary targets, which is not realistic. For a gas of identical molecules obeying the Maxwell-Boltzmann distribution, the average relative speed of two colliding molecules exceeds the mean molecular speed by a factor of Ö2. This correction results in l = 1/(Ö2)nσ, which implies: nlσ = 1/Ö2 » 0.707, not 1. Thus, the relation nlσ = 1 is an approximation—a pedagogical device for grasping the scaling behavior, but not the exact statistical result.  It is also known as the l-s relation, which expresses the inverse proportionality between the mean free path and the collision cross‑section: larger cross‑section, shorter path; smaller cross‑section, longer path. Its true value lies not in numerical precision but in the physical intuition it provides.

 

Real-World Nuance: It's Not Just Physical Size!

The l-σ relation provides a simple geometric picture of molecular collisions, allowing the concept of mean free path intuitively accessible. However, its simplicity comes at the cost of accuracy. Here are some key limitations.

1. Charged Particles: Attractive forces can pull particles together from distances far larger than their physical dimensions, effectively increasing the collision cross section, whereas repulsive forces can deflect particles before they come close, shrinking it.

2. Speed Dependence: At high relative speeds, particles zip past one another so quickly that they barely have time to feel the force—the effective cross‑section shrinks. At low speeds, they interact for longer periods, allowing attractive or repulsive forces to exert a greater influence and thereby modifying the collision probability.

3. Quantum Effects: At atomic and subatomic scales, the de Broglie wavelength of a particle can become effectively larger, diffraction and interference effects then alter the effective cross‑section in ways that have nothing to do with physical size.

Thus, while the geometric picture provides a useful starting point, the collision cross section is essentially a measure of interaction probability, shaped by interaction forces, relative motion, and, in many cases, quantum mechanics. Thus, the mean free path is ultimately determined not simply by geometry, but by the underlying physics governing particle interactions.

Key Takeaways:

1. Mean Free Path: The Average Distance Between Collisions

The mean free path is the average distance a particle travels between successive collisions. It is not the distance traveled by every particle, but rather the statistical average of many randomly distributed free-flight distances.

 

2. Collision Cross Section: The Effective Interaction Area

The collision cross section measures the probability that two particles will collide or interact. Although it has units of area, it should generally be interpreted as an effective interaction area rather than the particle’s “actual” geometric size. For ideal hard spheres, the cross-section is simply a geometric area, but for real molecules or ions, it depends on interaction force, relatively velocity, and quantum effects.

 

3. The l-σ Relation: Geometry Meets Probability

The formula nlσ = 1 means that, on average, a particle traveling a distance   sweeps out a cylindrical volume containing exactly one scattering target, so one collision is expected over that distance. This relation provides a geometric interpretation of the mean free path, l = 1/nσ, but it is an approximation that assumes stationary scatterers and neglects the relative motion of all particles in a real gas.

 

In short, the collision cross section tells us how large a target a particle presents, while the mean free path tells us how far a particle typically moves before hitting another. Increasing the number density of particles or enlarging their effective collision cross section increases the collision frequency and shortens the mean free path. This simple idea forms the microscopic foundation of diffusion, electrical conduction, viscosity, and many other transport phenomena.

 

A Real-Life Application: The Science of Fart Odor Transport The Source: Chemistry and Transport Physics The characteristic odor flatulence arises from volatile sulfur compounds, such as hydrogen sulfide and methanethiol. Although these compounds make up only a tiny fraction of the gas mixture, the human nose is remarkably sensitive to them. Once release, the odor molecules undergo a multi-stage process: Jet Advection – The gas is initially expelled as a turbulent jet, rapidly mixing with the surrounding air. Thermal convection –Released at about body temperature, the flatus may initially rise due to buoyancy. As it cools, this effect gradually diminishes.  Molecular diffusion: On a microscopic scale, individual odor molecules undergo random thermal motion, colliding with air molecules and slowly spreading the smell outward. Together, advection, convection, and diffusion result in how quickly the odor spreads.   Why Distance Matters There is no sharp boundary beyond which an odor suddenly vanishes—the concentration decreases continuously as the molecules disperse. The farther one stands from the source, the more the odor has been diluted by mixing with the air. Consequently, fewer odor molecules reach the olfactory receptors in the nose, reducing the probability that the smell exceeds the human detection threshold. Ventilation, air currents, room geometry, humidity, and temperature all affect the diffusion rate. In a poorly ventilated room, an odor may linger for several minutes; with good airflow, it dissipates in seconds.   The Probabilistic Nature of Smell Detecting an odor is fundamentally a probabilistic process. At high concentrations, a large number of odor molecules reach the nose every second, making detection almost certain. As the concentration falls, fewer molecules arrive, and the likelihood of detecting the smell decreases. Eventually the concentration drops below the detection threshold, and the odor becomes imperceptible, even though a small number of molecules may still be present.   Practical Takeaways To minimize significant olfactory exposure: Distance helps. Greater separation generally reduces odor concentration and therefore the probability of detection. Ventilation is even better. Fresh air rapidly dilutes and removes odor molecules from the room. Activated carbon masks are good at adsorbing sulfur compounds. Standard cloth, surgical, and N95 masks, however, are particulate filters—they capture aerosols, not gaseous molecules—and lack the necessary carbon bed to reduce odor in any meaningful way. If feasible, a small hand-held fan can be used to redirect the local airflow, spreading the odor away from your breathing zone and back toward its source.   Conclusion The next time a social situation requires a scientific description of “safety distance,” you can respond with a blend of statistical mechanics and social grace:  “Please maintain a radius of about three meters—or roughly two exponential decay lengths from the source—to guarantee negligible odor.” However, Feynman might have concluded it as follows: The molecules don’t know whether they’re carrying perfumes or unhealthy fumes—they simply obey the laws of physics.

 

The Moral of the Lesson: Mean Free Path and COVID‑19 The concept of mean free path provides a useful way of thinking about the spread of airborne diseases such as COVID-19. In this context, however, the idea appears in two different forms (Fernández-Terán et al., 2020). At the microscopic level, it describes how far virus-laden aerosols travel before colliding with another air molecule. At the macroscopic level, it serves as an analogy for the distance traveled by individuals in a crowd before encountering an infected person.   1. Safe Distancing: Dilution of Aerosol Concentration An infected individual continuously emits viral-laden aerosols through breathing, speaking, or sneezing. These aerosols are initially most concentrated near the source. As they spread through diffusion, gravitational settling, and ventilation, their concentration decreases (sometimes exponentially) with distance from the source. By increasing the physical separation between individuals, the aerosol concentration is effectively lowered. Statistically, a lower concentration means fewer virus-containing aerosols enter the breathing zone of an individual and therefore a lower probability of inhaling an infectious dose.   2. Wearing a Mask: Increasing the Probability of Collisions A mask is not a simple sieve. It is a complex three‑dimensional network of microscopic fibers that acts as a highly efficient particle filter. From the perspective of an incoming viral aerosol, the fibers form a dense array of collision targets. Without a mask: The air is equivalent to a low density of solid obstacles, allowing aerosols to move relatively unobstructed into the respiratory tract. With a mask: The probability of an aerosol encountering and adhering to a fiber increases dramatically. The result is a substantial reduction in the number of virus-laden aerosols that penetrate the filter and reach the lungs.   3. The Selection of Masks The effectiveness of a mask can be intuitively understood through the formula σnl = 1:   σ is the “effective” collision cross-section. n is the fiber number density (or density of collision targets). l is the mean free path. A useful analogy: increasing the number density of collision targets and their effective cross section reduces the distance a particle can move before an interaction occurs. This is why effective respirators (e.g., N95) fulfil the formula by combining a dense fiber network with suitable material to enhance aerosol capture. The goal is not to eliminate all aerosol motion but to maximize the chance that viral aerosols are intercepted before reaching the respiratory system.   Conclusion: COVID-19 transmission is not solely a problem for epidemiologists—it is also a problem in transport physics. By understanding the mean free path and collision cross-section, we can design effective interventions: safe distancing lengthens the total path particles must travel and masks reduce mean free path. Together, they lower the probability that viral aerosols complete their journey from one person to another.   A Note on COVID‑19 and Long-Term Health Risks Researchers continue to investigate the long-term consequences of SARS‑CoV‑2 infection, which extend beyond the acute respiratory phase to encompass potential effects on the cardiovascular, neurological, and respiratory systems (Tanrıverdi, 2024). Emerging evidence has also raised questions about possible links between COVID‑19 and cancer development. However, it is important to emphasize that current evidence remains preliminary, and no definitive causal relationship between COVID-19 and can cancer has been established. The field is still emerging, and further research is needed to clarify the underlying mechanisms and quantify long-term risks. This reinforces the importance of understanding aerosol transport: reducing exposure to viral droplets is not just about acute infection, but potentially about long‑term health consequences.

 

Richard P. Feynman To Yetta Farber, MARCH 30, 1982 Ms.Yetta Farber wrote to remind Feynman that she had once dated him at Cornell. She also had a story to tell that made her laugh every time she saw Feynman’s name in the papers. Immediately after their enjoyable date, word had gone out that there was a rapist, described as wearing “a brown or brown leather-like jacket,” loose on the Cornell campus. “A-hah! I said—I went out with this nice fellow and he wore a brown leather-like jacket! Maybe it was he. When you called me for another date, I said, ‘No, I’m busy.’” Ms. Farber had thought that Feynman was too young to be an assistant professor (in fact, Feynman was a full professor), so she was rather suspicious.   Dear Yetta, Naturally I could never understand why the girls I went out with in Ithaca wouldn’t go out with me again. At last I find out—it was my brown leather jacket! So often, was I thus frustrated by pretty girls (like you) that I came out to California. Since the weather was so much better I threw away my leather jacket and at last found someone who would go out with me more than once—so I married her.   I always thought that the girls in California were more tolerant—but now I know the inner workings of the phenomenon. Physics is much easier to understand.   Your former date, Richard P. Feynman (Source: Feynman, 2005, p. 345)   Feynman’s humorous letter about his brown leather jacket provides a valuable lesson in transport phenomena. The tragedy of Feynman’s dating life in Ithaca, New York, is not a charming anecdote—it is a real-world lesson in transport theory, molecular diffusion, and the physics of dating.   The Moral of the Lesson: Feynman’s Jacket Odor and the Mean Free Path 1. The Source: A Reservoir of Odor Molecules A well-worn leather jacket can act as a reservoir of volatile organic molecules (sweat, bacteria, and skin oils…). These molecules continuously evaporate from the jacket, creating a region of elevated odor around the jacket.   2. The Mean Free Path In short, the diffusion is slow and the mean free path is tiny. Although an odor molecule moves at hundreds of meters per second, they do not travel directly from the jacket to another person’s nose. It constantly collides with air molecules in a random walk. A molecule’s mean free path is typically less than a micrometer. Because it bounces backward, forward, and sideways continually, the molecule undergoes random walk, causing the smell to diffuse outward slowly.   3. Why Proximity Matters The concentration of odor molecules decreases with distance from the source. At a distance: When Feynman first met someone, he was standing far enough away that the flux of odor molecules reaching the girl’s nose was negligibly small. The "collision cross-section" of her olfactory receptors didn’t get enough molecules to trigger a warning. During a date: Once in close proximity—sitting across a table or walking side by side—the separation distance decreased and the exposure time increased. Under these conditions, enough odor molecules could diffuse to the olfactory receptors to confirm the jacket as a source.   A Reflection: The Hidden Variable The most amusing aspect of Feynman’s story is that he initially searched for the explanation in the wrong place. He thought the difference lay between the women of Ithaca and those of California. Eventually, he discovered that the “unknown” variable was neither geography nor human psychology, but his jacket. This is a lesson familiar to every physicist: when observations disagree with expectations, look for a hidden variable before constructing a new theory.   The Ultimate Takeaway In physics, as in dating, proximity changes everything. Thanks to a short mean free path, diffusion is inherently slow. A localized hazard—whether odor molecules, viral aerosols, or any other unwanted transport—can remain completely undetectable at a distance. The flux across a short distance could be surprisingly fast; you cannot rely on the sluggishness of diffusion to protect you. To lower the hazard rate of an undesirable transport process, do not rely on slow diffusion to save you—change the boundary conditions by eliminating the source at the boundary. And, for heaven’s sake, throw away the jacket or at least give it a generous spray of perfume.

 

Review Questions:

1. Explain why the stationary-target assumption is introduced when deriving the classical expression for the mean free path.

2. How would you explain the expression for the collision cross‑section of two different molecules?

3. How would you explain the physical meaning of the l-s relation and its limitations?

References:

Clausius, R. (1858). On the mean lengths of the paths described by the separate molecules of gaseous bodies. Philosophical Magazine, 15(101), 417–424.

Fernández-Terán, R., Sucre-Rosales, E., Echevarría, L., & Hernández, F. E. (2020). Social distancing during the COVID-19 pandemic: an analogy to explain collision cross-sections in chemical kinetics. Journal of Chemical Education, 97(12), 4540-4544.

Feynman, R. P. (2005). Perfectly reasonable deviations from the Beaten track: The letters of Richard P. Feynman (M. Feynman, ed.). New York: Basic Books.

Feynman, R. P., Leighton, R. B., & Sands, M. (1963). The Feynman Lectures on Physics, Vol I: Mainly mechanics, radiation, and heat. Reading, MA: Addison-Wesley.

Maxwell, J. C. (1867). On the dynamical theory of gases. Philosophical Transactions of the Royal Society of London, 157, pp. 49-88.

Tanrıverdi, Ö., Alkan, A., Karaoglu, T., Kitaplı, S., & Yildiz, A. (2024). COVID-19 and carcinogenesis: exploring the hidden links. Cureus, 16(8), e68303.