How do you read a code with its version label?

A code in ISIC Rev. 5, UN Statistics Division, should always be read together with its version label, because the label tells the reader which edition of the structure gives the code its meaning. The same four digits can point to different activities in Rev. 4 and Rev. 5, especially where categories were split, merged or moved. In a table, the version label belongs in the title, a footnote or a methods note, and in text it belongs beside the first mention of the code.

Where should the label sit so a later reader still finds it?

A version label does its work only where a later reader will actually see it. In a table, that means the title, a footnote, or a methods note that travels with the table when it is copied. In running text, it means beside the first mention of the code, not buried three pages later in an appendix a casual reader will never open. In a dataset, it means a field or a header that stays attached to the numbers through sorting, filtering and charting, rather than a note in a separate document that gets lost when the file is shared. The test is simple: take the code out of its original context, hand it to a stranger, and ask whether the stranger can still tell which edition of the structure gives it meaning. If the answer is no, the label has not yet been placed properly.

That placement matters because the same four digits can point to different activities in Rev. 4 and Rev. 5, especially where categories were split, merged or moved. A code without its label is therefore not a neutral fact waiting to be interpreted. It is an ambiguous string, and two careful readers, each assuming a different revision, can read it in two different ways and both believe they are right. Attaching the label, ISIC Rev. 5, UN Statistics Division, removes that ambiguity at the source, before the code enters a table, a chart or a quotation.

What changes when the same digits are read under Rev. 4?

A split means work that once sat in one category now sits in two or more, so a figure that once described a single category must now be read as several. A merge means work that once sat apart now sits together, so separate figures from the earlier revision combine into one. A move means work shifts from one branch of the hierarchy to another, so a figure that once belonged under one division now belongs under a different one. None of those changes is an error. Revisions are planned, documented and phased in precisely because the economy changes and the classification must keep pace. The error lies only in reading digits from one revision as if they belonged to another, without saying so.

For a writer, the practical consequence is that a code copied from an older publication cannot be carried forward on trust. The writer should open the UNSD Rev. 5 structure, find the corresponding category in the new revision, and check whether the scope is the same, narrower, broader, or moved. If the scope has changed, the writer should say so in the text or in a footnote, rather than letting the unchanged digits imply an unchanged meaning. Readers who are told about the change can adjust their interpretation. Readers who are not told will assume continuity that the revision has removed.

Which four details should travel with any quoted code?

Version label first: ISIC Rev. 5, UN Statistics Division, where Rev. 5 is meant, and never a different label silently substituted. Level second: whether the figure is a section, division, group or class figure, because a division total and a class figure answer different questions even when they share a word in their titles. Reference period third: the period the figure describes, so a reader can see whether a revision or an adoption date falls inside the series. Source fourth: the UNSD structure itself, or the national publication that applies it, so the definition behind the code can be checked rather than guessed from its title.

Keep those four details together in working notes, in methods notes and in table footnotes, and the code remains checkable long after the person who first wrote it has moved on. A code that travels with its label, its level, its period and its source is a small piece of infrastructure. A code that travels alone is a riddle, and riddles have no place in published statistics.

This guide describes how to read a labelled code in a statistical classification, and it does not decide anything for an individual reader. It does not determine a reader’s code, principal activity, or registration; for a particular record, confirmation should come from the national statistical institute or business register that holds the record.