We combine the photometric redshift data of Fernández-Soto et al. with the morphological data of Odewahn et al. for all galaxies with I<26.0 detected in the Hubble Deep Field. From this combined catalog we generate the morphological galaxy number counts and corresponding redshift distributions and compare these to the predictions of high-normalization zero- and passive-evolution models. From this comparison we conclude the following:

1. E/S0's are seen in numbers and over a redshift range consistent with zero-evolution or minimal passive-evolution to I=24. Beyond this limit, fewer E/S0's are observed than predicted implying a net negative evolutionary process—luminosity dimming, disassembly or masking by dust—at I>24. The breadth of the redshift distribution at faint magnitudes implies strong clustering or an extended epoch of formation commencing at z>3.

2. Spiral galaxies are present in numbers consistent with zero-evolution predictions to I=22. Beyond this magnitude some net positive evolution is required. Although the number counts are consistent with the passive-evolution predictions to I=26.0, the redshift distributions favor number and luminosity evolution, although few obvious mergers are seen (possibly classified as irregulars). We note that beyond z~2 very few ordered spirals are seen suggesting a formation epoch of spiral galaxies at z~1.5-2.

3. There is no obvious explanation for the late-type/irregular class, and this category requires further subdivision. While a small fraction of the population lies at low redshift (i.e., true irregulars), the majority lie at redshifts 1<z<3. At z>1.5 mergers are frequent and, taken in conjunction with the absence of normal spirals at z>2, the logical inference is that they represent the progenitors of normal spirals that form via hierarchical merging.