When $\rho = 0$ the distribution is symmetric about zero; thus one-sided lower critical values are $-1$ times the tabled one-sided upper critical values. Column headings are also labeled for the corresponding two-sided significance level and the percentage of the distribution less than the tabled value. $N$ is the number of observations; the degrees of freedom is two less than this.

Table 1: Critical values of the bivariate normal sample correlation coefficient $\rho$.

	Percent
	90	95	97.5	99	99.5	99.9	99.95
	One-sided $\alpha$
	.10	.05	.025	.01	.005	.001	.0005
	Two-sided $\alpha$
	.20	.10	.05	.02	.01	.002	.001
$N$
3	.951	.988	.997	1.000	1.000	1.000	1.000
4	.800	.900	.950	.980	.990	.998	.999
5	.687	.805	.878	.934	.959	.986	.991
6	.608	.729	.811	.882	.917	.963	.974
7	.551	.669	.755	.833	.875	.935	.951
8	.507	.622	.707	.789	.834	.905	.925
9	.472	.582	.666	.750	.798	.875	.898
10	.443	.549	.632	.716	.765	.847	.872
11	.419	.522	.602	.685	.735	.820	.847
12	.398	.497	.576	.658	.708	.795	.823
13	.380	.476	.553	.634	.684	.772	.801
14	.365	.458	.533	.612	.661	.750	.780
15	.351	.441	.514	.592	.641	.730	.760
16	.338	.426	.497	.574	.623	.711	.742
17	.327	.412	.482	.558	.606	.694	.725
18	.317	.400	.468	.543	.590	.678	.708
19	.308	.389	.456	.529	.575	.662	.693
20	.299	.378	.444	.516	.562	.648	.679
25	.265	.337	.396	.462	.505	.588	.618
30	.241	.306	.361	.423	.463	.542	.570
35	.222	.283	.334	.392	.430	.505	.532
40	.207	.264	.312	.367	.403	.474	.501
45	.195	.248	.294	.346	.380	.449	.474
50	.184	.235	.279	.328	.361	.427	.451
55	.176	.224	.266	.313	.345	.408	.432
60	.168	.214	.254	.300	.330	.391	.414
65	.161	.206	.244	.288	.317	.376	.399
70	.155	.198	.235	.278	.306	.363	.385
75	.150	.191	.227	.268	.296	.351	.372
80	.145	.185	.220	.260	.286	.341	.361
85	.140	.180	.213	.252	.278	.331	.351
90	.136	.175	.207	.245	.270	.322	.341
95	.133	.170	.202	.238	.263	.313	.332
100	.129	.165	.197	.232	.257	.305	.324