Below are arrayed the denominational categories used in constructing the two indexes of religious diversity employed in this study's empirical analyses of church membership in 1890 and 1906. With rare exceptions, the original categorizations of the Census statisticians have been preserved, and the original labels for their categories retained. When classifications were rearranged, the primary motivation was to ensure the comparability of religious designations over time. Decisions on which groups to shift were thus guided principally by historical considerations, such as the existence of a common parent body or the shared ethnicity of combined memberships.

Because some data, particularly the data for 1890, were not minutely disaggregated, several obvious errors (for example, the inclusion in 1890 of members of the Reformed Catholic Church with Latin rite and Orthodox faithful) could not be corrected. For this same reason, members of some smaller Protestant bodies (e.g., Moravians) are counted in 1890 as “Other” rather than as “Protestant.” The gross numbers involved in these arithmetic misallocations, however, are small; they should not sway any results.

The Protestant-Catholic-Jewish-Other diversity index relies, as its name implies, on just four broad categories of religious affiliation. They are noted in CAPITAL letters at the left margin, and all groups named to the right and below are subsumed under those labels. The divisions of importance when diversity among Protestants is calculated are indicated by italics. Seventy-one distinct Protestant bodies are represented in the sample of cities in 1890. For the diversity index, they were collapsed into 21 categories, representing larger denominational “families.” The 127 Protestant bodies reporting in 1906 were similarly fit into 26 general classifications.