FileGizmo glossary
Transpose
Transposing a table swaps its two axes. The first row becomes the first column, the second row becomes the second column, and a value that sat in row two column seven ends up in row seven column two. Applied twice it returns the original table exactly.
A table has two axes, and which one holds the records is a layout decision rather than a fact about the data. Transposing swaps them.
The rule is simple enough to state exactly: the value at row n, column m moves to row m, column n. A table of 3 rows and 40 columns becomes one of 40 rows and 3 columns. Nothing is added, nothing is removed, and every value keeps the same neighbours it had, viewed from the other direction.
Why it is reversible
Transposing twice returns the original. That property is worth knowing because it is the fastest way to test whether a tool has done the operation or something adjacent to it.
The usual way a tool breaks it is by treating the header row as metadata rather than as a row. If the headers are held aside during the flip and then written back across the top, the result is not a transpose, and running it twice gives back something different from what went in. A header row is only a header by convention; to the operation it is the first row.
Where it is used
The common case is an export that was written to be read rather than to be imported. Financial and management reports put periods across the top and metrics down the side, because that is how people scan them. Almost every import format wants the opposite: one record per row, one field per column.
The other case is charting. Plotting libraries have opinions about which axis holds the series, and transposing is usually faster than restating the question.
The practical limit
Spreadsheets are not symmetric about their two axes. Most will open a file with a million rows and stop at a few thousand columns.
That means transposing a tall file is technically correct and practically useless: a ten thousand row export becomes a ten thousand column file that opens in almost nothing. Where the goal is a summary rather than a genuine flip, grouping the data is usually the operation actually wanted.
Frequently asked questions
Does transposing lose data?
No. Every value keeps its neighbours, only along the other axis. Transposing twice gives back exactly the file you started with, which is the simplest way to check that a tool did it properly.
What happens to the header row?
It becomes the first column. Treating it as special would break the reversibility, so a correct transpose moves it like any other row.
Why would a file need transposing?
Usually because it was laid out for reading and now has to be imported. Reports tend to put periods across the top; most import formats want one record per row.