Table 1. A comparison of codon usage among sequences of influenza A HA, NA, and NP genes.

	Miyata	Hamming
HA vs. NP	0.998**	1.000**
HA vs. NA	0.997**	0.987*
HA+ vs. NP	0.999**	1.000**
HA+ vs. NA	0.999**	1.000**
HA- vs. NP	0.706	0.211
HA- vs. NA	0.692	0.053
NA vs. NP	0.390	0.841

We write X vs. Y to denote that the codon usage in gene X was compared to a null distribution of 10,000 Monte Carlo trials generated from usage in gene Y. (We do not count a tie as one of the 10,000 trials.) We report the proportion of these Monte Carlo trials for which the volatility was exceeded by the observed volatility of X in the Miyata or Hamming metric. HA⁺ denotes 130 epitopic residues of HA1; HA^- denotes the remaining 199 nonepitopic residues. A single asterisk indicates statistical significance for a two-tailed test at the 5% confidence level; a double asterisk indicates significance at the 1% confidence level. Note that the codon usage of HA, especially in the epitopic residues, is significantly biased toward diversification relative to the usage of NA and NP, but the codon usage of the nonepitopic residues of HA is not significantly biased relative to NA or NP, nor is NA biased relative to NP