AbstractModels in colour engineering can have implicit customary usage, outside of whose parameters one is not supposed to venture. CIELAB is one such model: although able to accommodate a prediction of asymmetric matches, CIELAB is used in practice to compare colours only for a given illuminant/observer condition. Should we therefore deem CIELAB to be an infinity of spaces, one for each illuminant/observer combination? This question has its roots in an oddity of CIELAB’s chromatic‐adaptation model. A chromatic‐adaptation oddity also entered CIE models that emerged after CIELAB, and leads to a similarly perplexing question.