Answer: (ii) Logical paths in a program

Cyclomatic complexity (McCabe, 1976) is a software metric that measures the number of linearly independent paths through a program's control-flow graph — i.e., the number of logical decision paths (loops, conditionals) that must be tested. It is computed as

$$V(G) = E - N + 2$$

where $E$ = number of edges, $N$ = number of nodes of the control-flow graph (equivalently $V(G) = P + 1$, where $P$ = number of predicate/decision nodes).

Its main uses:

• Determines the minimum number of test cases required to achieve branch/path coverage.
• Indicates the complexity/testability of a module — high complexity means more paths, more testing, and higher defect potential.

Why the others are wrong:

• (i) Lines of code — that is a size measure (LOC), not cyclomatic complexity.
• (iii) Execution speed — a performance concern, unrelated.
• (iv) Code documentation quality — a maintainability/documentation concern, unrelated.

So cyclomatic complexity primarily measures logical paths in a program — option (ii).