Table 1. Reliability of several constant stimuli protocols based on simulated administration to the 590 normosmic distribution, n = 100 subjects with 500 replications.

	Number of Stimuli	16	12	8	6	(6)x2 a
Evenly Distributed	Dilutions Administered	1–16	1–3, 5–7, 9–11, 13–16	2–16 (Even) b	2, 5, 8, 9, 12, 15	(2, 5, 8, 9, 12, 15)x2
	Mean Reliability (SE) c	0.843 (0.001)	0.794 (0.002)	0.706 (0.003)	0.649 (0.003)	0.805 (0.002)
	5%, 50%, 95%	0.789, 0.844, 0.889	0.720, 0.797, 0.850	0.607, 0.711, 0.787	0.538, 0.657, 0.741	0.739, 0.809, 0.859
	% of Convergence Failures	0.2%	2.6%	29.8%	24.6%	0.6%
Centered	Dilutions Administered	NA d	5–16	7–14	8–13	(8–13)x2
	Mean Reliability (SE)		0.805 (0.002)	0.732 (0.002)	0.660 (0.002)	0.776 (0.002)
	5%, 50%, 95%		0.739, 0.807, 0.863	0.657, 0.736, 0.801	0.568, 0.668, 0.736	0.712, 0.781, 0.827
	% of Convergence Failures		1.4%	1.2%	8.2%	0.0%
Tails	Dilutions Administered	NA	1–6, 11–16	5–8, 13–16	5–7, 14–16	(5–7, 14–16)x2
	Mean Reliability (SE)		0.719 (0.003)	0.684 (0.002)	0.551 (0.003)	0.723 (0.002)
	5%, 50%, 95%		0.616, 0.725, 0.802	0.586, 0.687, 0.770	0.435, 0.550, 0.660	0.645, 0.726, 0.800
	% of Convergence Failures		6.0%	11.8%	29.2%	1.2%
Shifted	Dilutions Administered	NA	1–12	1, 3, 5, 7–11	1, 3, 5, 7, 9, 11	(1, 3, 5, 7, 9, 11)x2
	Mean Reliability (SE)		0.795 (0.002)	0.732 (0.002)	0.642 (0.002)	0.783 (0.002)
	5%, 50%, 95%		0.735, 0.798, 0.853	0.658, 0.735, 0.798	0.551, 0.645, 0.730	0.715, 0.789, 0.836
	% of Convergence Failures		2.4%	10.0%	29.2%	0.8%

^aThis configuration presents the identical 6 dilutions used for a given protocol twice for a total of 12 stimuli

^bA configuration using 8 odd dilutions yielded similar results

^cSE = standard error

^dNA = not applicable.