...
The deposited dataset must include sufficient description to understand and reuse it.
If the data consists of one or two fields/variables, it may be enough to provide the description of those fields in the abstract, but generally each dataset should have at least one Read Me file (aka data key or data dictionary). The Read Me can be in a separate plain text or delimited tabular file, or can be part of a tabular data file, e.g., a new tab in an .xslx document.
More info on creating ReadMe files can be found on the Libraries' website https://library.si.edu/research/describing-your-data-data-dictionaries
The Read Me should include (where applicable):
...
Any datasets which do not have either an abstract or a separate ReadMe file that includes sufficient information to interpret the deposited dataset are not guaranteed to be permanently managed by central SI services after the owner of the data has retired or moved on.More info on creating ReadMe files can be found on the Libraries' website https://library.si.edu/research/describing-your-data-data-dictionaries