Characteristic Speeds: Mean, RMS, and Most Probable

The Maxwell-Boltzmann speed distribution is a "lopsided bell curve", a skewed peak with a long high-speed tail to be precise, and a lopsided curve has no single "the speed." Three natural summaries emerge, and they are genuinely different numbers: the most probable speed $v_p$ (where the curve peaks), the mean speed $\bar v$ (the ordinary average), and the root-mean-square speed $v_{\text{rms}}$ (the square root of the average of $v^2$ ). For a gas at temperature $T$ , they are $v_p=\sqrt{2k_BT/m}$ , $\bar v=\sqrt{8k_BT/\pi m}$ , and $v_{\text{rms}}=\sqrt{3k_BT/m}$ — three constants times the same thermal velocity scale $\sqrt{k_BT/m}$ . What makes these three speeds a favourite exam topic — and a genuinely useful piece of physics — is that their ratios are universal. The temperature and mass cancel entirely: $v_p:\bar v:v_{\text{rms}}=\sqrt 2:\sqrt{8/\pi}:\sqrt 3\approx1.414:1.596:1.732$ . So regardless of which gas or how hot, the mean speed is always $2/\sqrt\pi\approx1.128$ times the most probable, and the rms is always $\sqrt{3/2}\approx1.225$ times the most probable. The ordering $v_p<\bar v<v_{\text{rms}}$ never changes: the peak sits below the mean because the distribution has a long high-speed tail that drags the average up, and the rms sits above the mean because squaring weights the fast molecules even more. Each speed is the right one for a different physical question. The most probable speed sets the peak of Doppler-broadened spectral lines. The mean speed enters the collision frequency and mean free path. The rms speed is fixed by the kinetic energy, $\tfrac12 m v_{\text{rms}}^2=\tfrac32 k_BT$ , and is the one that connects directly to temperature. Knowing which is which — and that they differ by fixed factors near unity — is the difference between an order-of-magnitude estimate and a correct one.

Key takeaways

A molecular gas at temperature $T$ has three characteristic speeds: most probable $v_p=\sqrt{2k_BT/m}$ , mean $\bar v=\sqrt{8k_BT/\pi m}$ , and root-mean-square $v_{\text{rms}}=\sqrt{3k_BT/m}$ .
They always obey $v_p<\bar v<v_{\text{rms}}$ : the peak is below the mean (long high-speed tail), the rms is above the mean (squaring favours fast molecules).
Their ratios are universal, independent of gas and temperature: $\bar v/v_p=2/\sqrt\pi\approx1.128$ , $v_{\text{rms}}/v_p=\sqrt{3/2}\approx1.225$ , $v_{\text{rms}}/\bar v=\sqrt{3\pi/8}\approx1.085$ .
All three scale as $\sqrt{T/m}$ , so lighter molecules and hotter gases are faster; helium moves $\sqrt{m_{N_2}/m_{He}}\approx2.65$ times faster than nitrogen at the same temperature.
The rms speed is fixed by the mean kinetic energy $\tfrac12 m v_{\text{rms}}^2=\tfrac32 k_BT$ , tying it directly to temperature.
Key idea: the three characteristic speeds are the same thermal scale $\sqrt{k_BT/m}$ times fixed constants $\sqrt2,\sqrt{8/\pi},\sqrt3$ , so their ordering and ratios are universal.

The core insight

The most probable, mean, and root-mean-square speeds of a gas are $v_p=\sqrt{2k_BT/m}$ , $\bar v=\sqrt{8k_BT/\pi m}$ , $v_{\text{rms}}=\sqrt{3k_BT/m}$ . They always satisfy $v_p<\bar v<v_{\text{rms}}$ , and their ratios ( $2/\sqrt\pi$ , $\sqrt{3/2}$ ) are independent of gas and temperature.

Where this lives today

The characteristic speeds are working numbers across physics and chemistry. In spectroscopy, the Doppler width of an emission line is set by $v_p$ (or $\bar v$ ) along the line of sight, so measuring a line profile yields a gas temperature — the standard "Doppler thermometer" in plasmas, flames, and the interstellar medium. In vacuum technology and surface science, the mean speed $\bar v$ fixes the effusion rate and the flux of molecules onto a surface, governing deposition and pump-down times. In chemical kinetics, the high-speed tail beyond $v_p$ supplies the molecules energetic enough to react, so reaction rates are exquisitely sensitive to the tail. And in isotope separation, the mass scaling $v\propto1/\sqrt m$ — the reason gaseous UF $_6$ of different isotopes effuses at slightly different rates — is the physical basis of gaseous-diffusion and centrifuge enrichment. Molecular-dynamics thermostats target these speeds to set simulation temperatures.

The physical picture

The speed distribution $f(v)$ starts at zero (no molecule has zero speed once the shell factor $4\pi v^2$ is included), rises to a peak, then decays with a long tail toward high speeds. Because the tail is long and one-sided, the three ways of summarizing the curve give three different answers, and they line up in a fixed order. The peak — the most probable speed — comes first. The ordinary mean is pulled to the right of the peak by the tail. The rms is pulled even further right, because forming $\langle v^2\rangle$ weights the fast molecules by their speed squared. The payoff is that these are not three arbitrary numbers but three landmarks on one universal shape. Rescale the horizontal axis by $\sqrt{k_BT/m}$ and every gas at every temperature collapses onto the same master curve, with $v_p$ , $\bar v$ , and $v_{\text{rms}}$ sitting at the fixed dimensionless positions $\sqrt2$ , $\sqrt{8/\pi}$ , $\sqrt3$ . Change the gas or the temperature and the whole curve stretches or shrinks, but the three landmarks keep their fixed relative spacing.

Figure

The Maxwell-Boltzmann speed distribution f(v) rises from zero, peaks at the most probable speed v_p, then decays with a long high-speed tail. The three speeds line up in fixed order.

The second figure is the punchline: overlay the same curve for two different gases at the same temperature. The lighter gas (helium) is a broad, low curve reaching to high speeds; the heavier gas (nitrogen) is a narrow, tall curve concentrated at low speeds. The three landmarks move together — helium's $v_p$ , $\bar v$ , and $v_{\text{rms}}$ are all $\sqrt{m_{N_2}/m_{He}}\approx2.65$ times nitrogen's — but the ratios among the three within each gas are identical.

Figure

Punchline: at the same temperature, lighter helium (broad, fast) and heavier nitrogen (narrow, slow) have curves that differ by the mass factor √(m_N2/m_He) ≈ 2.65, yet the internal ratios v_p:v̄:v_rms are identical for both.

The core equations

All three speeds are moments of the Maxwell-Boltzmann speed distribution $f(v)=4\pi(m/2\pi k_BT)^{3/2}v^2e^{-mv^2/2k_BT}$ . The most probable speed is where $f(v)$ peaks; setting $df/dv=0$ gives it directly.

v_p = \sqrt{\frac{2k_BT}{m}}\qquad\left(\frac{df}{dv}\bigg|_{v_p}=0\right) \qquad (1)

where: $v_p$ — the most probable speed, the location of the peak of $f(v)$ ; $m$ — molecular mass; $T$ — temperature. It is the speed a randomly chosen molecule is most likely to have. The mean speed is the first moment, $\bar v=\int_0^\infty v f(v)\,dv$ , and evaluates to a clean closed form.

\bar v = \int_0^\infty v\,f(v)\,dv = \sqrt{\frac{8k_BT}{\pi m}} \qquad (2)

where: $\bar v$ — the arithmetic mean speed, the average of $|\mathbf v|$ over the gas. It appears in kinetic-theory expressions for collision frequency, mean free path, and effusion flux. The root-mean-square speed is the square root of the second moment, and it is fixed by the mean kinetic energy through equipartition.

v_{\text{rms}} = \sqrt{\langle v^2\rangle} = \sqrt{\frac{3k_BT}{m}},\qquad \tfrac12 m v_{\text{rms}}^2 = \tfrac32 k_BT \qquad (3)

where: $\langle v^2\rangle=3k_BT/m$ is the mean-square speed; the right-hand identity is equipartition — three translational degrees of freedom, each contributing $\tfrac12 k_BT$ . This is the speed that ties directly to temperature. The universal content is the set of ratios, in which $T$ and $m$ cancel completely.

v_p:\bar v:v_{\text{rms}} = \sqrt 2:\sqrt{\tfrac 8\pi}:\sqrt 3,\qquad \frac{\bar v}{v_p}=\frac{2}{\sqrt\pi},\quad \frac{v_{\text{rms}}}{v_p}=\sqrt{\tfrac32} \qquad (4)

where: the numerical values are $\sqrt2\approx1.414$ , $\sqrt{8/\pi}\approx1.596$ , $\sqrt3\approx1.732$ , giving $\bar v/v_p\approx1.128$ and $v_{\text{rms}}/v_p\approx1.225$ . These ratios hold for every gas at every temperature.

Derivation AThe three speeds from the moments of the speed distribution

The goal is to derive all three characteristic speeds — most probable, mean, and root-mean-square — as specific moments of the Maxwell-Boltzmann speed distribution, and to obtain their exact closed forms $v_p=\sqrt{2k_BT/m}$ , $\bar v=\sqrt{8k_BT/\pi m}$ , $v_{\text{rms}}=\sqrt{3k_BT/m}$ . The strategy is to locate the peak by differentiation for $v_p$ , and to evaluate the first and second moments using the standard Gamma-function integrals $\int_0^\infty v^n e^{-\alpha v^2}\,dv$ .

Step 1. Most probable speed. Set $df/dv=0$ with $f(v)\propto v^2 e^{-mv^2/2k_BT}$ . Differentiating the product, $\frac{d}{dv}(v^2 e^{-\alpha v^2})=(2v-2\alpha v^3)e^{-\alpha v^2}=0$ with $\alpha=m/2k_BT$ , so $v^2=1/\alpha=2k_BT/m$ , giving $v_p=\sqrt{2k_BT/m}$ .

Step 2. Mean speed. The first moment is $\bar v=\int_0^\infty v\,f(v)\,dv=4\pi(m/2\pi k_BT)^{3/2}\int_0^\infty v^3 e^{-\alpha v^2}\,dv$ . Using $\int_0^\infty v^3 e^{-\alpha v^2}\,dv=1/(2\alpha^2)$ , the algebra collapses to $\bar v=\sqrt{8k_BT/\pi m}$ .

Step 3. Root-mean-square speed. The second moment is $\langle v^2\rangle=4\pi(m/2\pi k_BT)^{3/2}\int_0^\infty v^4 e^{-\alpha v^2}\,dv$ . Using $\int_0^\infty v^4 e^{-\alpha v^2}\,dv=\frac{3}{8}\sqrt{\pi/\alpha^5}$ , this gives $\langle v^2\rangle=3k_BT/m$ , so $v_{\text{rms}}=\sqrt{3k_BT/m}$ . Why they differ. The peak $v_p$ ignores the tail; the mean $\bar v$ weights by $v$ ; the rms weights by $v^2$ . Each higher power of $v$ shifts weight toward the fast molecules, moving the summary speed to larger values — hence $v_p<\bar v<v_{\text{rms}}$ .

Step 4. Equipartition check on $v_{\text{rms}}$ : $\tfrac12 m\langle v^2\rangle=\tfrac12 m(3k_BT/m)=\tfrac32 k_BT$ , the mean kinetic energy of three translational degrees of freedom — the rms speed is the one fixed by temperature.

Go deeper · Derivation B · 1 minThe universal ordering and ratios

Derivation BThe universal ordering and ratios

The goal is to prove that the ordering $v_p<\bar v<v_{\text{rms}}$ holds for every gas at every temperature and that the ratios between the three speeds are pure numbers, independent of $m$ and $T$ . The strategy is to factor the common thermal scale $\sqrt{k_BT/m}$ out of all three, leaving only the dimensionless coefficients $\sqrt2$ , $\sqrt{8/\pi}$ , $\sqrt3$ , and to compare these coefficients directly.

Step 1. Write each speed as a coefficient times the thermal scale $u\equiv\sqrt{k_BT/m}$ : $v_p=\sqrt2\,u$ , $\bar v=\sqrt{8/\pi}\,u$ , $v_{\text{rms}}=\sqrt3\,u$ . The scale $u$ carries all the mass and temperature dependence; the coefficients are pure numbers.

Step 2. Compare coefficients numerically: $\sqrt2=1.4142$ , $\sqrt{8/\pi}=1.5958$ , $\sqrt3=1.7321$ . Since $1.4142<1.5958<1.7321$ , the ordering $v_p<\bar v<v_{\text{rms}}$ holds identically, whatever $u$ is.

Step 3. Form the ratios: $\bar v/v_p=\sqrt{(8/\pi)/2}=\sqrt{4/\pi}=2/\sqrt\pi\approx1.128$ , and $v_{\text{rms}}/v_p=\sqrt{3/2}\approx1.225$ , and $v_{\text{rms}}/\bar v=\sqrt{3/(8/\pi)}=\sqrt{3\pi/8}\approx1.085$ . The thermal scale $u$ cancels in every ratio, so these are universal constants. Why the mean sits below the rms. By the Cauchy-Schwarz (or Jensen) inequality, $\bar v=\langle v\rangle\le\sqrt{\langle v^2\rangle}=v_{\text{rms}}$ , with equality only for a delta-function distribution. Because the speed distribution has genuine spread, $\text{Var}(v)>0$ and the inequality is strict: $\bar v<v_{\text{rms}}$ always.

Step 4. Consequence for gas comparisons: at fixed $T$ , all three speeds scale as $1/\sqrt m$ , so the ratio of speeds of two gases is $\sqrt{m_2/m_1}$ . Helium versus nitrogen gives $\sqrt{m_{N_2}/m_{He}}=\sqrt{28.01/4.003}=2.646$ — helium is $2.65$ times faster at every landmark.

Limiting cases

The three speeds are proportional to $\sqrt{T/m}$ , so their behaviour in limits follows the single thermal scale — but the ordering and ratios never change.

Case	Result	Meaning
High temperature	all three $\propto\sqrt T$ grow	Faster molecules, but $v_p:\bar v:v_{\text{rms}}$ unchanged.
Heavy molecule	all three $\propto1/\sqrt m$ shrink	SF $_6$ is much slower than H $_2$ ; ratios identical.
$T\to0$ (classical)	all three $\to0$	Motion freezes; quantum statistics intervene before this.
Comparing two gases at same $T$	$v_1/v_2=\sqrt{m_2/m_1}$	Lighter is faster; the basis of gaseous diffusion isotope separation.
Delta-function limit (no spread)	$v_p=\bar v=v_{\text{rms}}$	Only if all molecules had one speed — never true for a real gas.

Common trapCommon pitfall

The three speeds are close (within about $22\%$ of each other) but not interchangeable, and using the wrong one gives systematically wrong answers. The rms speed $v_{\text{rms}}=\sqrt{3k_BT/m}$ is the one fixed by kinetic energy and temperature — use it for energy arguments. The mean speed $\bar v=\sqrt{8k_BT/\pi m}$ is the one that enters collision frequency, mean free path, and effusion — use it for transport. The most probable speed $v_p=\sqrt{2k_BT/m}$ locates the peak of the distribution and Doppler line centres. A frequent error is to compute "the speed" as $\sqrt{3k_BT/m}$ and then plug it into a mean-free-path formula that actually calls for $\bar v$ , overestimating by the factor $v_{\text{rms}}/\bar v\approx1.09$ . Always ask which moment the physics wants.

Worked examples

Example 1 — All three speeds for nitrogen at room temperature (enumerate, apply, boundary number). Take N $_2$ ( $M=28.013\,\mathrm{g/mol}$ , $m=28.013\times1.66054\times10^{-27}=4.652\times10^{-26}\,\mathrm{kg}$ ) at $T=300\,\mathrm{K}$ .

Step 1. Thermal scale $u=\sqrt{k_BT/m}=\sqrt{(1.380649\times10^{-23})(300)/(4.652\times10^{-26})}=\sqrt{8.90\times10^{4}}=298.4\,\mathrm{m/s}$ .

Step 2. Most probable: $v_p=\sqrt2\,u=1.414\times298.4=\mathbf{422\,m/s}$ .

Step 3. Mean: $\bar v=\sqrt{8/\pi}\,u=1.596\times298.4=\mathbf{476\,m/s}$ . RMS: $v_{\text{rms}}=\sqrt3\,u=1.732\times298.4=\mathbf{517\,m/s}$ .

Step 4. Boundary check on the ratios: $\bar v/v_p=476/422=1.128=2/\sqrt\pi$ ✓, and $v_{\text{rms}}/v_p=517/422=1.225=\sqrt{3/2}$ ✓. The three landmarks obey the universal ratios exactly — a good internal consistency test.

Example 2 — Comparing helium and nitrogen; and the rms of helium.

Step 1. At the same $T$ , the speed ratio of two gases is $v(He)/v(N_2)=\sqrt{m_{N_2}/m_{He}}=\sqrt{28.013/4.0026}=\sqrt{6.999}=\mathbf{2.646}$ . Helium molecules move about $2.65$ times faster than nitrogen at any temperature.

Step 2. Compute helium's rms directly: $m_{He}=4.0026\times1.66054\times10^{-27}=6.647\times10^{-27}\,\mathrm{kg}$ , so $v_{\text{rms}}=\sqrt{3(1.380649\times10^{-23})(300)/(6.647\times10^{-27})}=\sqrt{1.869\times10^{6}}=\mathbf{1367\,m/s}$ .

Step 3. Cross-check with the ratio: nitrogen's $v_{\text{rms}}=517\,\mathrm{m/s}$ (Example 1) times $2.646$ gives $517\times2.646=1368\,\mathrm{m/s}$ — matching the direct calculation. This mass scaling is exactly why lighter gases effuse and diffuse faster, the principle behind gaseous uranium-isotope enrichment.

Origins · 1 minmost probable speed and how this was discovered

most probable speed

statistical-mechanics concept · Maxwell 1860

The most probable speed $v_p=\sqrt{2k_BT/m}$ is the speed at which the Maxwell-Boltzmann speed distribution peaks — the single most likely value of $|\mathbf v|$ for a molecule drawn at random from a gas at temperature $T$ . It is the natural unit of molecular speed: expressed in units of $v_p$ , the whole distribution takes a gas-independent shape, and the mean and rms speeds become the fixed multiples $2/\sqrt\pi$ and $\sqrt{3/2}$ . Because the peak sets the centre of Doppler-broadened spectral lines and marks where most molecules cluster, $v_p$ is the reference speed against which the mean and rms are measured, and the anchor of the universal ordering $v_p<\bar v<v_{\text{rms}}$ .

History

1859

— Maxwell presents the speed distribution and identifies the peak (most probable) speed.

1860

— Maxwell's "Illustrations of the Dynamical Theory of Gases" gives the full speed law and its moments.

1873

— Maxwell and others connect the mean speed to the mean free path and transport coefficients.

1920

— Otto Stern measures molecular speeds directly with a rotating-drum molecular beam, confirming the distribution.

1955

— Miller and Kusch's precision molecular-beam experiment verifies the Maxwell speed distribution to high accuracy.

Going further

References. Pathria & Beale, Statistical Mechanics, 3rd ed., Ch. 6 (kinetic theory and the Maxwell distribution) · Reif, Fundamentals of Statistical and Thermal Physics, §7.10 · Blundell & Blundell, Concepts in Thermal Physics, §5–6 · Kardar, Statistical Physics of Particles, §3–4 · Atkins, Physical Chemistry, Ch. on molecular motion. The three speeds $v_p=\sqrt{2k_BT/m}$ , $\bar v=\sqrt{8k_BT/\pi m}$ , $v_{\text{rms}}=\sqrt{3k_BT/m}$ follow from: (i) $df/dv=0$ giving $v_p=\sqrt{2k_BT/m}$ ; (ii) $\int_0^\infty v f\,dv=\sqrt{8k_BT/\pi m}$ and $\int_0^\infty v^2 f\,dv=3k_BT/m$ via the Gamma integrals; (iii) the ratios $\bar v/v_p=2/\sqrt\pi\approx1.128$ , $v_{\text{rms}}/v_p=\sqrt{3/2}\approx1.225$ , $v_{\text{rms}}/\bar v=\sqrt{3\pi/8}\approx1.085$ ; (iv) the numeric speeds for N $_2$ at 300 K ( $422,476,517\,\mathrm{m/s}$ ) and $v_{\text{rms}}(He)=1367\,\mathrm{m/s}$ ; (v) the mass scaling $v(He)/v(N_2)=2.646$ . Constants: $k_B=1.380649\times10^{-23}\,\mathrm{J/K}$ .

Builds on: Maxwell-Boltzmann Speed Distribution · Maxwell-Boltzmann Velocity Distribution · Classical Ideal Gas Leads to: Equipartition Theorem · Virial Theorem in Statistical Mechanics · Thermal de Broglie Wavelength

Why are there three different "average" speeds?

Because the speed distribution is asymmetric — a bell curve with a long high-speed tail — so there is no unique way to summarize it with one number. The most probable speed is the peak, the mean is the ordinary arithmetic average, and the rms is the square root of the average of $v^2$ . Each is a different moment of the same distribution, and each answers a different physical question. For a symmetric distribution they would coincide, but the one-sided tail of the speed distribution separates them.

Which speed should I use in a given problem?

Match the moment to the physics. Use $v_{\text{rms}}$ for anything involving kinetic energy or pressure, since $\tfrac12 mv_{\text{rms}}^2=\tfrac32 k_BT$ . Use $\bar v$ for transport quantities — collision frequency, mean free path, effusion flux — because these average the speed linearly. Use $v_p$ to locate the peak of the distribution or the centre of a Doppler-broadened line. Using the wrong one introduces a systematic error of up to about $22\%$ .

Why is v_p the smallest of the three?

The most probable speed marks the peak of the distribution, but the distribution has a long tail extending to high speeds and no tail below zero. The tail drags the mean above the peak, and the rms — which weights by $v^2$ and so favours fast molecules even more — sits above the mean. So the ordering $v_p<\bar v<v_{\text{rms}}$ reflects the increasing sensitivity of each moment to the high-speed tail.

Do the ratios really not depend on the gas?

Correct — and this is the elegant part. Every speed is $\sqrt{k_BT/m}$ times a pure number ( $\sqrt2$ , $\sqrt{8/\pi}$ , or $\sqrt3$ ), so when you form a ratio the $\sqrt{k_BT/m}$ cancels entirely. The ratios $\bar v/v_p=2/\sqrt\pi$ and $v_{\text{rms}}/v_p=\sqrt{3/2}$ are the same for helium, nitrogen, or uranium hexafluoride, at any temperature. Only the overall scale changes with gas and temperature.

How much faster is a light gas than a heavy one?

At the same temperature, all three speeds scale as $1/\sqrt m$ , so the ratio of speeds equals the square root of the inverse mass ratio, $v_1/v_2=\sqrt{m_2/m_1}$ . Helium ( $4\,\mathrm{u}$ ) versus nitrogen ( $28\,\mathrm{u}$ ) gives $\sqrt{28/4}\approx2.65$ . Hydrogen versus oxygen gives $\sqrt{32/2}=4$ . This is why hydrogen and helium escape planetary atmospheres and why gaseous diffusion can separate isotopes despite their tiny mass difference.

How were these speeds first measured?

Otto Stern pioneered molecular-beam methods in 1920: molecules leaving an oven pass through a rotating apparatus, and the rotation sorts them by speed (fast molecules travel further before the drum turns), reconstructing the distribution. Later precision experiments by Miller and Kusch (1955) confirmed the Maxwell-Boltzmann speed distribution and its characteristic speeds to high accuracy. Modern time-of-flight and Doppler techniques measure them routinely.

Is the mean speed the same as the mean velocity?

No — this is a crucial distinction. The mean velocity $\langle\mathbf v\rangle$ is zero for a gas at rest, because velocity is a vector and opposite directions cancel. The mean speed $\bar v$ is the average of the magnitude $|\mathbf v|$ , which is always positive and equals $\sqrt{8k_BT/\pi m}$ . A gas can have zero mean velocity (no bulk flow) while every molecule has a large speed.