See: https://www.cambridge.org/core/journals/journal-of-functional-programming/article/genuine-sieve-of-eratosthenes/FD3E90871269020CA6C64C25AB8A4FBD

The current version of SICP JS claims in 3.5.2 (as did many, prior to this paper) that the algorithm given there faithfully represents the original Sieve of Eratosthenes. The paper above disproves this.

SICP JS should reflect the unfaithfulness.