Stop a sound in a hard room and it does not stop. It bounces, losing a little at each reflection, and the time it takes to fall by 60 decibels — a millionth of its energy — is the reverberation time, $T_{60}$. In a plant room it is the difference between an instruction heard and an instruction guessed at.

Wallace Sabine measured it in 1900 and got a relation that has not needed changing: $T_{60} = \dfrac{0.161\,V}{A}$ — T-sixty equals nought point one-six-one V over A. $V$ is the room volume in cubic metres, and $A$ is the total absorption in square metres of sabins — not an area of surface, but an area of perfect absorber that would swallow as much. The 0.161 is metric and carries the speed of sound and the 60 dB decay inside it; in cubic feet and square feet the same physics wears 0.049, and mixing the two is a factor-of-three error that looks entirely plausible.

Where does $A$ come from? Surface by surface: $A = S_1\alpha_1 + S_2\alpha_2 + S_3\alpha_3$, where each $S$ is a surface area in m² and each $\alpha$ — alpha — is that surface's absorption coefficient, a bare number from 0 (a perfect mirror) to 1 (an open window, which absorbs everything by letting it leave). Sealed concrete is about 0.02; an acoustic ceiling tile is 0.70 or better. Areas matter as much as coefficients: a large mediocre surface routinely beats a small excellent one.

Sabine has an edge, and it is worth knowing where. His derivation assumes each reflection loses only a sliver of the energy — true in a live room, false in a dead one. Push it to a heavily treated studio and it runs long by a third or more; push it to a room with no walls at all and it still refuses to return zero. Eyring fixes that: $T_{60} = \dfrac{0.161\,V}{-S\,\ln(1-\bar{\alpha})}$, where $S$ is the total surface area of every boundary and $\bar{\alpha}$ — alpha-bar — is the area-weighted average coefficient. The logarithm is natural, not base ten, and it acts on $1-\bar{\alpha}$: the fraction of energy that survives a reflection. Rule of thumb — below $\bar{\alpha} \approx 0.2$ the two agree closely; above it, use Eyring.

One rating, one warning. $\mathrm{NRC} = \dfrac{\alpha_{250} + \alpha_{500} + \alpha_{1000} + \alpha_{2000}}{4}$ is a material's four-band average, rounded to the nearest 0.05 — the single figure that goes on a submittal. Look at what it hides: the four bands are averaged, so a panel useless at 250 Hz and superb at 2 kHz scores the same as one flat across the range. And 125 Hz is not in the list at all, which is precisely where a fan and a diesel do their work.