Table 3. Bias in the estimation of λ(n,n) of the island, direct and censored direct methods, and their relative speeds.

n	Bias (%)			Speed ratio
	Island	Direct	Censored	Island:direct	Island:censored
2400	+0.9	–0.2	–0.2	60	60
1200	+0.7	0.0	0.0	14	14
800	+0.8	+0.4	+0.4	6	6
600	+0.6	+0.5	+0.5	3	3
480	+0.5	+0.8	+0.6	2	2
400	+0.4	+0.8	+0.3	1.4	1.5
343	+0.4	+1.2	+0.4	1.0	1.2
300	+0.3	+1.4	+0.5	0.8	1.1
267	+0.1	+1.4	+0.3	0.6	1.0
240	0.0	+1.6	+0.1	0.5	0.9
218	–0.1	+1.4	–0.1	0.4	0.8
200	0.0	+1.6	–0.1	0.3	0.8

Maximum-likelihood estimates λ were derived using the island, direct and censored methods from n × n sequence comparisons. For the island method with cutoff score c = 28, approximately 10⁶ scores were generated, yielding a standard error of 0.1%. For the direct method, approximately 6.1 × 10⁵ scores were generated, yielding a standard error of 0.1%. For the censored method, sufficient scores above the threshold c = 28 were generated to yield a standard error of 0.1% (23); this number ranged from 6.1 × 10⁵ for n = 2400 to 8.4 × 10⁵ for n = 200. Biases are calculated assuming the correct values for λ(n,n) are those of the points plotted in Figure 4. Speed ratios are the computation times required, respectively, by the direct and censored methods divided by the time required by the island method; high speed ratios favor the island method.