Companion data and code for: "The repetition ratio r/R measures length and encoding, not linguistic status: a reanalysis of Sproat (2014)"

The repetition ratio rule (RRR), based on the ratio r/R of geminate to total repeats, has been proposed as a discriminant between linguistic and non-linguistic symbol systems. We show that for short and moderate-length inscriptions the expected r/R (precisely, the ratio of expectations E[r]/E[R]) under a uniform-random null is 2/L, where L is the perinscription token count. Contents: manuscript (PDF + LaTeX source), reproducibility script (compute_rR.py, Python 3 standard library only run python3 compute_rR.py to regenerate every numeric result), figure-generation scripts, and source corpora: SSA and DVV top-100 name lists, 58 commodity labels, 10,767 CDLI Mesopotamian seal legends (derived from github.com/cdli-gh/data, CC attribution), SCOWL word list, and the ICIT doubled-sign-code mapping. See README.md.