A queueing identity stating that for a stable system observed over the long run, the long‑run average number L of items in the system equals the long‑run average effective arrival rate λ multiplied by the average time W an item spends in the system: L = λW; 'stable' means long‑run arrival and departure rates are equal and averages are well‑defined.

Principle

Little’s Law is a conservation‑of‑flow invariance: average inventory equals average throughput times average time in system. It holds regardless of arrival and service time distributions, service order, or internal routing, provided the system is stable and measurements use consistent long‑run averages.

Demonstration

Illustrative scenario → A buffer ahead of a machine has long‑run throughput λ = 10 parts/hour. Measured average time in system W = 0.5 hour. Little’s Law predicts L = λW = 10 × 0.5 = 5 parts average in the buffer. Recognition: if measured L differs from λW the discrepancy indicates measurement error, non‑stability or misdefinition of the counted system.

Misapplication

Applying L = λW to transient periods, to non‑conservative systems where items are created/destroyed within the counted boundary, or using instantaneous rates and times instead of long‑run averages. The semantic error is treating momentary snapshots or non‑stationary arrivals as if the long‑run identity applies without checking stability and consistent counting.

Consequence

Correct use: permits direct computation of one of {L, λ, W} when the other two are known and supports capacity sizing, WIP targets and lead‑time policies. Misuse: yields incorrect inventory or lead‑time targets and misguided operational decisions. Mechanistically, inconsistent measurement windows or instability (λ ≠ departure rate) break the relation and invalidate derived planning parameters.

Reversal

Little’s Law does not apply when flow conservation fails (e.g., batching that alters the counted 'unit', processes that merge/split items without consistent accounting), when arrivals and departures are not balanced over the measurement horizon, or when averages are non‑ergodic. In multiclass networks it applies by class only with careful state definitions or via generalized network formulations.

Boundary

Clearly within: a stable production stage or buffer where discrete items arrive and depart and long‑run averages exist. Boundary case: heavily time‑varying demand where averaging windows must be chosen with care. Clearly outside: instantaneous queue length during a transient ramp‑up, or continuous flows where 'item' identity is ambiguous without discretization.

Semantic Tension

Little’s Law (aggregate conservation) can conflict with detailed stochastic queueing models that are required to evaluate variance, tail delays or service‑level quantiles; the tension is between using an aggregate identity for planning and using distributional models for reliability guarantees.

Synthesis

Little’s Law provides a robust, distribution‑free bridge between throughput, work‑in‑process and lead time for stable systems; its practical power is enabling measurement‑based control of WIP/lead time trade‑offs while recognizing it supplies no information about variability or transient behaviour.