How Do I Handle Spelling Variants?

What to check first in this record

Handle spelling variants by searching every variant you can find, including phonetic spellings and initials, and by matching on number and unit rather than spelling alone. Records follow the spelling on the form, not the family spelling.

The strongest clue is often the least glamorous: a number, a date, a place. Glamorous clues, badges and stories, point in many directions at once. Begin with the filters the catalogue itself uses, because a search that fights the catalogue structure will return noise however carefully it is worded.

Making sense of what you find

Abbreviations are the easiest place to go wrong, because the same letters can stand for different things in different periods and different services. A date written in one format on one sheet and another format on the next is the same date wearing two formats. Convert carefully, and keep the original. Cross-references inside a file, see another sheet, another number, another unit, are the file telling you where its own gaps are filled. Follow them.

Where this kind of research usually stalls

A photograph with no caption, a medal with no roll entry, and a story with no dates are all the same problem: a clue without a filter. If the dates you have are approximate, say so, and search a range rather than a single year. A record filed under a neighbouring year is a common miss. A common surname in a large unit needs a second filter before any search is meaningful. Without one, you are sampling the unit, not finding a person.

The order to work in

Before the first step, gather every clue you already hold into one place. Steps taken with scattered clues produce scattered results. If a step produces nothing, record that nothing. A documented negative result is part of the research, and it stops the same step being repeated blindly.

1. Name the clue exactly: what it is, what it says, and what it does not say.
2. Choose the first record that clue points to, and note its reference before opening it.
3. Extract only dated facts from that record, and add them to a first timeline with source notes.
4. Let the first timeline choose the second record, and keep guesses in a separate list until a record joins them.

What to open after this one

Do not start a new cluster until this one has yielded its dated facts. Half-finished clusters produce timelines that mix two record types badly. The end of a guide is a good place to review the timeline for unreferenced entries. Remove them, or mark them, before they harden into facts. A research project ends, properly, not with a flourish but with a referenced timeline and a list of questions still open.