How did the section count change to 22?

The section count in ISIC Rev. 5, UN Statistics Division, reached 22 sections, labelled A to V, partly because the old Section J was split into two sections, J and K. That split added a section letter at the top of the hierarchy and pushed later sections to new letters. A reader comparing section level tables across revisions must therefore remap section letters, not assume that a letter means the same block of activities in both revisions.

What happened to the letter J?

The clearest way into the section count is through the single split named in the opening paragraph. In Rev. 4, Section J covered a broad block of activities. In Rev. 5, that block became two sections, J and K. One letter therefore became two, and the total rose by one at that point alone. Everything else about the section count follows from that kind of arithmetic: new letters appear where a block was divided, and the count records the outcome rather than explaining it.

The split also explains why the rest of the alphabet shifted. Sections are labelled in order, so inserting a new section partway through the sequence pushes every later section onto a new letter. A section that sat under one letter in Rev. 4 may sit under a different letter in Rev. 5, even when its own scope changed little. The letter is an address, not a definition, and addresses change when a new house is built in the middle of the street.

Why is remapping letters more urgent than it sounds?

Readers remember letters. A policy note, a spreadsheet column, or a chart legend may carry a section letter with no version label beside it, and a later reader may read it as if letters were stable across revisions. After the J and K split, that reading fails. The same letter can point to a different block of activities in each revision, and a different letter can point to a block that barely changed. Any comparison that matches letters instead of matching definitions will mix unlike things together and then present the mixture as a trend.

The remedy is unglamorous and effective. Before comparing section level tables, write out the Rev. 4 letter and the Rev. 5 letter for each block in the comparison, side by side, and check each pair against the UNSD structure. Where a letter changed but the scope did not, note the relabelling. Where the scope changed as well, note that too, because the two kinds of change need different footnotes. A table that shows its remapping invites checking. A table that hides it invites error.

How should a reader handle the sections after K?

Sections beyond the split deserve the same care as the split itself, because the push to new letters affects all of them at once. A reader scanning a Rev. 5 table from A to V will see 22 sections in order, and the sequence will look as natural as any alphabet. That natural look is misleading if the reader carries Rev. 4 expectations into it. The right approach is to read the Rev. 5 sequence on its own terms first: what does each section cover, where does it start, and where does it stop? Only then should the reader ask how each section relates to its Rev. 4 predecessor.

Short sections and long sections both need this treatment. Length says nothing about stability. A section that kept its relative position in the list may still have gained or lost scope, and a section that moved letters may have kept its scope almost intact. Position, letter, and scope are three separate facts, and a careful reader keeps them separate instead of letting one stand in for the others.

A small exercise with two tables

Take a section level table from Rev. 4 and the matching table from Rev. 5, and try to line them up without looking at the correspondence material. The mismatch appears fast: letters do not line up, totals do not reconcile, and at least one block seems to have moved. Now repeat the exercise with the correspondence material open, matching blocks by their definitions rather than by their letters. The second attempt is slower, and it is the only one that produces a comparison worth publishing.

This exercise also shows why the opening paragraph insists on remapping rather than assuming. Assumption feels efficient because it skips the mapping step, but the time saved is repaid later, with interest, when a chart must be corrected or a paragraph withdrawn. Mapping first is the cheaper order of work, even though it does not feel like it at the time.

Nothing here decides a code, a principal activity, or a registration outcome for any individual reader or business. This is general information about a statistical classification only, and where a specific record is concerned, the national statistical institute or business register that holds that record is the place to confirm it.