Spearman's own model of a single general factor and rival sampling-theory models, which explain the same positive correlations among test scores as an artifact of overlapping mental processes rather than one underlying capacity, are statistically indistinguishable from the correlational data alone.
What would resolve this Evidence that does not depend on the same correlational test-score structure both rival models already fit equally well, for example converging biological or neurological findings for a single general capacity distinct from a bundle of overlapping specific processes.
OpenWikipedia: G Factor (Psychometrics)