We discuss two tests of the hypothesis that the first genes crc re assembled from exons, The hypothesis of exon shuffling in the progenote predicts that intron phases will be correlated so that exons will be an integer number of codons and predicts that the exons will he correlated with compact regions of polypeptide chain, These predictions have been tested on ancient conserved proteins (proteins without introns in prokaryotes hut with introns in eukaryotes) and hold with high statistical significance. We conclude that introns are correlated with computer features of proteins 15-, 22-, or 30-amino acid residues long, as was predicted by ''The Exon Theory of Genes.''