The eight tiers were fixed before any of these three instruments was opened. A 2010 survey of thirty-seven US educator-performers, a 2019 log of 2,038 calls at 160 London sessions, and fourteen degree programs' published repertoire lists, retrieved in August 2026. None cites the others. All three fall off at the same place.
Share of each tier's seated titles that the instrument names. The top two tiers are the instrumental core; below them the agreement thins, which is the shape the tier definitions predict and the reason the core is stated as a narrow claim.
Two departures from the descent, stated here rather than left to be found. At Index 1B the survey reaches 94.3% where the log reaches 73.6%, which is the tier the two instruments most disagree about and the reason 1B is printed as core-adjacent rather than core. At Index 5 all three rise again: it is the vocal division, measured on a different question, and its reach figures are not a claim about instrumental call.
Instrument reach is read off TABLE VI.3.b: the survey column is Distilled plus Sub-Floor plus Reconciled, the log column is Logged plus Reconciled. Panel figures are App. H. Index 5 is 5A, 5B and 5C combined, 227 titles.
The same log, read as a distribution rather than as a checklist. Each group's share of the Index's 729 seated titles against its share of the 2,038 calls the log recorded. The two shares invert.
Read deeper into the log and the gradient holds: Index 1 holds nine of the raw log's ten most-called titles, twenty-two of its top twenty-five, forty-three of its top fifty and seventy-six of its top hundred. Those count the log as kept; the Top 25 figure in the last panel counts the strata its keeper published, which is a different set of twenty-five titles. Index 2's call share is the residual of the two figures the volume prints, since the three groups partition all 729 titles.
Titles seated from each successive band of two hundred ranks in the jazzstandards.com thousand. The screening runs that produced the proposals never saw the rank column, which is what makes this the strongest external check in the volume, and the series is recomputable by a reader from the published ranking and the declines at App. C.
Take two indexes, pair every title on the first with every title on the second, and ask how often the measure puts the second index's title ahead. Ties count as half. Near zero the measure agrees with the volume. At fifty it cannot see a difference between the two tiers. Above fifty it reverses them. The recording ranking built the pool and drew no tier line, so this is a held-out test rather than a circular one.
Rows run in the order the volume prints them, by what the ranking returns. The two call instruments stay left of the line on every row. The ranking crosses it three times and runs to 85 at Index 3 against Index 4, where it prefers the older, longer-in-print material the period house exists to hold. On that row the log and the survey are near fifty because both are largely silent there, which is a tie rather than an opposition, and the volume says so. Figures for Index 3 against Index 4 are printed as approximate in the source.
Starting from what each compiler put at the top of his own apparatus and asking where those titles land here.