Boltzmann's entropy S = k_B ln(Ω): Visual walkthrough

Step 1 — What is a "way it can happen"? (the microstate)

WHAT. Picture four little boxes in a row. Each box holds one coin, and each coin shows either Heads (H) or Tails (T). One complete filling-in of all four boxes — say H T T H — is one exact snapshot. We call each such complete snapshot a microstate.

WHY. Before we can count ways, we must agree on what "one way" is. A microstate is the smallest, fully-detailed unit of counting: every coin pinned down, nothing left vague.

PICTURE. Look at figure s01. The amber row is one microstate — every coin decided. The faint cyan rows below are other microstates: change even one coin and it becomes a different snapshot.

Figure — Boltzmann's entropy S = k_B ln(Ω)

Step 2 — Grouping snapshots that "look the same" (the macrostate)

WHAT. From outside you can't see individual coins — you can only read a bulk number, say "how many Heads total." So we sort every microstate into bins labelled by that bulk number: 0 heads, 1 head, 2 heads, ... A bin is a macrostate.

WHY. Real measurements are coarse. You measure energy $E$ , volume $V$ , particle number $N$ — never the exact microstate. So the physically meaningful object is the bin, not the snapshot. The question "how likely is this bin?" becomes "how many snapshots landed in it?"

PICTURE. In figure s02 every snapshot (cyan dot) drops into a bin by its head-count. The bin "2 heads" is visibly the fattest — more dots fell there. That fatness is the whole story.

Term-by-term: $Ω$ = the count; a microstate = one dot; a macrostate = one bin. Bigger bin ⇒ bigger $Ω$ ⇒ (we will see) more entropy.

Step 3 — Joining two systems: counts multiply

WHAT. Put system $A$ (with $Ω_{A}$ snapshots) beside an independent system $B$ (with $Ω_{B}$ snapshots). For each choice of $A$ 's snapshot you may pair any of $B$ 's. So the total number of joint snapshots is the product.

$Ω_{A B} = Ω_{A} \times Ω_{B}$

Reading it term-by-term:

$Ω_{A}$ = ways for the left system alone,
$Ω_{B}$ = ways for the right system alone,
$Ω_{A B}$ = ways for the two treated as one combined system,
the $\times$ appears because the choices are independent — nothing about $A$ restricts $B$ .

WHY THIS TOOL (multiplication, not addition). Whenever two choices are made independently, the totals multiply — exactly why two dice give $6 \times 6 = 36$ outcomes, not $6 + 6$ . We are not assuming physics here, only the arithmetic of independent choices.

PICTURE. Figure s03 is a grid: $A$ 's options run along the rows, $B$ 's along the columns. Every cell is one joint microstate. Count the cells — it's rows $\times$ columns. The picture is the multiplication.

Step 4 — What we demand of entropy: it must add

WHAT. We want a quantity $S$ — "entropy" — that we will define from $Ω$ somehow, $S = f (Ω)$ . Thermodynamics already treats entropy as extensive: glue two identical blocks together and the entropy doubles, just like mass or energy. So we insist:

$S_{A B} = S_{A} + S_{B}$

Term-by-term: $S_{A}$ = entropy of the left block, $S_{B}$ = of the right, $S_{A B}$ = of the combined block, and the $+$ is the property we are imposing by hand to match real thermodynamics.

WHY. This is the pivotal design choice. Nature gave us products in Step 3; thermodynamics wants sums here. Our job is to find the one function $f$ that bridges the two.

PICTURE. Figure s04 stacks the tension side by side: on the left, counts combine by $\times$ ; on the right, our required entropy combines by $+$ . The double arrow marks the gap a function must close.

Step 5 — The one function that turns $\times$ into $+$ : the logarithm

WHAT. Combine Steps 3 and 4. Since $S = f (Ω)$ and $Ω_{A B} = Ω_{A} Ω_{B}$ , additivity demands

$f (Ω_{A} Ω_{B}) = f (Ω_{A}) + f (Ω_{B}) .$

We solve this. Differentiate both sides with respect to $Ω_{A}$ (call it $x$ , and $Ω_{B} = y$ ): $y f^{'} (x y) = f^{'} (x) .$ Set $x = 1$ : $y f^{'} (y) = f^{'} (1) \equiv k$ , a constant. Hence $f^{'} (y) = \frac{k}{y} ⟹ f (y) = k ln y + C .$ And $C = 0$ because one way means zero entropy: $f (1) = 0$ (if there's only a single snapshot, you have no missing information at all).

$f (Ω) = k ln Ω$

WHY THE LOGARITHM (why this tool and no other). The logarithm is defined by the identity $ln (x y) = ln x + ln y$ . It is the unique continuous function that swallows a product and spits out a sum. Our whole tension — nature multiplies, thermodynamics adds — has exactly one continuous cure, and its name is $ln$ .

PICTURE. Figure s05 shows it happening: two counts on a multiplying number-line map, through the $ln$ curve, onto an adding number-line. Watch $Ω_{A} Ω_{B}$ on the bottom axis land at $ln Ω_{A} + ln Ω_{B}$ on the side axis.

Step 6 — Fixing the constant: $k = k_{B}$

WHAT. The functional equation left $k$ free. We fix it by matching one known case — an ideal gas — to measured thermodynamic entropy. The match forces $k$ to be Boltzmann's constant:

$S = k_{B} ln Ω, k_{B} = 1.380649 \times 1 0^{- 23} J/K .$

WHY. $ln Ω$ is a pure number, but real entropy is measured in joules-per-kelvin. $k_{B}$ is the exchange rate that converts "number of ways" into thermodynamic units — and it is exactly the constant that makes $1/ T = \partial S / \partial E$ come out in kelvin.

PICTURE. Figure s06: the bare curve $ln Ω$ (cyan) is rescaled by the tiny amber factor $k_{B}$ into the physical entropy $S$ (amber). Same shape, physical units.

Step 7 — The degenerate case: $Ω = 1$ (Third Law)

WHAT. What if a system has only one possible microstate? A perfect crystal at absolute zero sits frozen in its single lowest-energy arrangement, so $Ω = 1$ . Then

$S = k_{B} ln 1 = k_{B} \times 0 = 0.$

WHY. $ln 1 = 0$ because you multiply nothing to get $1$ — no choices remain, no hidden information. Boltzmann's counting formula predicts the Third Law of Thermodynamics: entropy vanishes when only one way is left. This is not an add-on; it falls straight out of the log.

PICTURE. Figure s07: as the number of available snapshots collapses to a single dot, the entropy bar shrinks smoothly to the floor $S = 0$ .

Step 8 — The other edge: entropy can fall as energy rises

WHAT. We usually think "more energy ⇒ more entropy." But $S$ tracks $ln Ω$ , not $E$ . In a capped-energy system — spins that can only point up or down in a magnetic field — pumping in energy eventually reduces the number of arrangements (all spins forced to align), so $Ω$ falls and $S$ falls. There $\partial S / \partial E < 0$ , the hallmark of negative temperature.

WHY. This edge case guards against the false shortcut "entropy = energy." Boltzmann's formula only ever counts ways; if the ways run out at high energy, the entropy drops. Covering this keeps our picture honest across all regimes, not just the everyday heating one.

PICTURE. Figure s08 plots $Ω$ against energy for a two-state spin system: it rises, peaks at the middle energy, then falls back — a hill, not a ramp. The entropy follows the same hill.

The one-picture summary

Figure s09 compresses the whole walkthrough into a single flow: snapshots → bin them → join systems (counts multiply) → demand entropy adds → the only bridge is $ln$ → scale by $k_{B}$ → read off the edge cases ( $Ω = 1$ gives $S = 0$ ; capped systems can lose entropy).

Recall Feynman: the whole walk in plain words

Start with coins. One exact filling of all the coins is a snapshot. From across the room you can't see the coins — you can only count "how many heads," so you sort snapshots into bins. Some bins hold way more snapshots than others; that headcount of a bin is $Ω$ .

Now push two coin-tables together. For every snapshot on the left you can pair any snapshot on the right, so the joint count multiplies — like a grid whose cells are rows times columns. But we want our "entropy" number to behave like weight: two tables should give twice the entropy, adding, not multiplying.

There's exactly one math gadget that eats a multiply and gives back an add: the logarithm, because $ln (ab) = ln a + ln b$ . So entropy has to be $ln$ of the count. Solving the little equation confirms it and forces the leftover constant to vanish when there's only one way ( $ln 1 = 0$ ). Finally we sprinkle in $k_{B}$ , a tiny number that turns "count of ways" into real joules-per-kelvin.

Two sanity checks close the loop: a perfect crystal frozen into its single arrangement has $Ω = 1$ , so $S = 0$ — the Third Law, for free. And in a spin system where the ways run out at high energy, entropy can actually drop as you add energy — proof that entropy counts ways, never energy itself.

Recall Quick self-test

Why do counts multiply but entropies add? ::: Joining independent systems pairs every microstate of one with every microstate of the other (grid of cells = rows × columns), so counts multiply; we demand entropy be extensive so it adds — the $ln$ reconciles the two. What forces $f (Ω) = k ln Ω$ and not some other function? ::: $ln$ is the unique continuous solution of $f (x y) = f (x) + f (y)$ . What does $Ω = 1$ give and which law is that? ::: $S = k_{B} ln 1 = 0$ , the Third Law. When can adding energy lower entropy? ::: In capped-energy systems (e.g. spins) where $Ω$ decreases at high $E$ — negative-temperature regime.

Connections

Parent: Boltzmann's entropy
Microcanonical Ensemble
Gibbs Entropy $S=-k_B\sum p_i\ln p_i$
Temperature as $\partial S/\partial E$
Third Law of Thermodynamics
Negative Temperature & Spin Systems
Gibbs Paradox & Indistinguishability
Shannon Information Entropy
Entropy and the Second Law of Thermodynamics