What Is Entropy? Boltzmann's Idea Explained Simply

Entropy is one of the most famous and most misunderstood quantities in physics. It is often described as disorder, which is a useful picture but not a precise one. A more accurate idea is that entropy counts how many ways a system can be arranged on the small scale while still looking the same on the large scale. Systems tend to move toward states with more of these arrangements because there are simply more of them.

This post builds the idea from coin tosses, introduces Ludwig Boltzmann's formula, computes entropy changes for a melting ice cube and a reversible heat transfer, and corrects some common mistakes.

Macrostates and microstates

A macrostate is a description of a system in measurable terms, such as its temperature, volume and total energy. A microstate is a complete description of every particle's position and velocity. Many microstates can produce the same macrostate.

Coins make this concrete. Toss 10 coins and look at the number of heads. Each specific sequence of heads and tails is one microstate. The macrostate "exactly 5 heads" is compatible with many more microstates than the macrostate "all heads."

Boltzmann's formula

Ludwig Boltzmann proposed in the 1870s that entropy is proportional to the logarithm of the number of microstates, W, that match a macrostate:

S = k_B ln W

The constant k_B is the Boltzmann constant, 1.380649 × 10⁻²³ J/K. The logarithm is needed so that entropies of independent systems add while their microstate counts multiply. If you combine two systems with W₁ and W₂ microstates, the total is W₁ × W₂, and the logarithm turns that product into a sum.

Boltzmann's formula is engraved on his gravestone in Vienna. It connects the thermodynamic quantity S to the microscopic counting of particle arrangements.

The coin table

The table below uses the coin example. The number of microstates for a macrostate with h heads out of N coins is the binomial coefficient C(N, h), and the entropy is k_B ln C(N, h) in units of k_B.

Coins (N)	Heads (h)	Microstates W	ln W	S / k_B
10	0	1	0	0
10	5	252	5.53	5.53
10	10	1	0	0
100	50	about 1.0 × 10²⁹	66.8	66.8

Most of the outcomes cluster near half heads. That is why a random shuffle almost always ends up looking mixed, and why it almost never returns to all heads or all tails by chance. The same statistics applies to molecules in a box. The equal-split state is overwhelmingly more probable simply because it has more arrangements.

Entropy change in a reversible process

Classical thermodynamics defines entropy change through heat. For a reversible process at constant temperature T, the entropy change is:

ΔS = Q_rev / T

A worked example: 1000 J of heat flows reversibly into a large reservoir at 300 K. The entropy change is ΔS = 1000 / 300 = 3.33 J/K.

Boltzmann's counting and this heat-based definition agree. Adding heat to a system at a given temperature increases its energy and therefore the number of accessible microstates.

Melting ice

Melting ice is a clean test. At 0 °C (273.15 K), ice and liquid water coexist, and melting requires a latent heat of fusion of about 6.01 kJ per mole of water.

1. Heat per mole: Q = 6010 J/mol.
2. Temperature: T = 273.15 K.
3. Entropy change: ΔS = Q / T = 6010 / 273.15 ≈ 22.0 J/(mol·K).

The liquid has more accessible arrangements than the crystal, because molecules can move past each other rather than sitting in fixed lattice positions. So the entropy of the water rises during melting.

Entropy in the second law

The second law says the total entropy of an isolated system does not decrease. Heat moving from a hot body to a cold body increases total entropy, because the hot body loses entropy at a higher temperature than the cold body gains it at a lower one. The net change is positive, so the process runs spontaneously in that direction.

A refrigerator lowers the entropy of its interior, but only by dumping more entropy into the kitchen. The total always rises. Life also follows this rule. Organisms build low-entropy structures by consuming food and releasing heat and waste, which carries more entropy into the surroundings.

Common misconceptions

Entropy is not just disorder. A shuffled deck is not messier in any visible way than a sorted one, yet the shuffled state corresponds to vastly more card orders. Describing entropy as "mess" can mislead because the count of arrangements is the real quantity.

Entropy does not always increase in the everyday sense. A crystal forming from solution can decrease the entropy of the solute while the total entropy, including the released heat, still increases. Looking at only one part of the system hides the full picture.

Entropy is not energy. Adding energy to a system usually raises entropy, but entropy is a count of arrangements, not a reservoir of energy.

Why gases spread out

Open a bottle of perfume in one corner of a room and the scent reaches the whole room within minutes. Nothing pushes the molecules across the room. Each molecule moves randomly, and the number of arrangements in which molecules fill the whole room vastly exceeds the number in which they stay bunched near the bottle. The spread-out state is simply the one that is overwhelmingly likely.

The same counting explains why a gas in a box never spontaneously gathers into one half. The probability of finding all of a gas's molecules in one half of a box is about two raised to the minus N power, where N is the number of molecules. For a few molecules this is an everyday chance. For a mole of gas, around 10²³ molecules, it is far too small to ever observe. Entropy is what makes such one-way behaviour look certain at everyday scales.

Entropy and information

Claude Shannon later defined information entropy using the same mathematical form, which measures uncertainty in a message. The two entropies are related conceptually and share the logarithmic counting idea, though they use different units and scales. Connecting them is an active area of research, especially in quantum information and the thermodynamics of computation.

Key takeaways

Entropy counts microstates, and Boltzmann's formula S = k_B ln W turns that count into a thermodynamic quantity. Systems move toward macrostates with more microstates because those states are overwhelmingly likely. In a reversible process, entropy change is heat divided by temperature. Melting, heat flow and refrigeration all follow the same rule that total entropy of an isolated system never decreases.