How to Distinguish a Surname Variant from a Transcription Error

What the question asks, and what it does not

When a researcher finds the same family written two different ways, the instinct is to ask whether the second form is “the real name” or “a mistake.” That framing hides the actual question. A surname can appear in more than one spelling for two quite different reasons: the person who created the record wrote it that way, or someone who copied the record later changed it. The first is a recorded spelling variant. The second is a transcription or indexing error. Distinguishing them is a question about evidence, not about which form looks more familiar.

This guide deals with that narrow decision. It does not attempt to establish a surname’s etymology, and it does not assume that two similar names belong to the same family. The examples come from a United States indexing context, because the source consulted here — FamilySearch’s guide to name variations in United States indexes and records — is written for that setting. The underlying principle applies wherever records have been copied, abstracted, or indexed: the closer you get to the original document, the more reliable your answer.

Three things that get confused

Beginners often use one word, “variant,” for three separate situations. Keeping them apart is the first practical step.

• A recorded spelling variant. The original document itself shows the different spelling. The clerk, priest, enumerator, or the person signing wrote it that way. Spelling was not standardised in the way it is today, and the same individual could be recorded differently at different times.
• A transcription or indexing error. The original document shows one spelling, but an indexer, transcriber, abstractor, or optical character recognition process rendered it as another. FamilySearch’s guide describes two kinds of error in this area: the creator of the record may have misspelled the name in the original, or the transcriber or indexer may have misread the original or mistyped the entry.
• A merely similar name. A different surname that happens to look or sound close. Similarity is not evidence of a relationship, and merging two people because their names sound alike is one of the most common beginner mistakes.

Only the first of these is a variant in the strict sense. The second is an artefact of the copy you are reading. The third is a separate surname until records show otherwise.

Why the distinction matters

The distinction changes what you do next. If a spelling difference is an indexing error, the fix is a search strategy: try the form the original is likely to contain, and record that you did so. If it is a genuine recorded variant, the fix is to add that form to your understanding of how the family was written in a particular place and period — and to keep looking for it in other records. If it is a different surname, the correct action is to stop and not merge the two people.

Treating an index error as a family variant can send you searching for a name that never existed in the records. Treating a genuine variant as an error can make you dismiss a document that actually belongs to your family. Both mistakes come from deciding too early, before the original has been checked.

What supports a variant reading

FamilySearch’s guide lists a series of ways a name may have been written or misread, including initials and abbreviations, the use of a middle name instead of a first name, different vowels, added or dropped double letters, transposed letters, misread letters, and phonetic substitutions. Read carefully, these are search suggestions, not findings. That the name RADCLIFFE might be indexed as RATCLIFFE or RADKLIFFE tells you where to look; it does not tell you what any particular record says.

What actually supports calling a form a variant is evidence from the document itself:

• The original image, or a photocopy of it, shows the different spelling.
• The variant recurs in records of the same place and period, especially records created by the same clerk or office.
• The variant is consistent with the handwriting of the period — for example, letters that are commonly confused in the script used.
• The same person can be identified across the records by other details, such as dates, places, relatives, occupation, or a distinctive combination of given names.

No single one of these is conclusive. The strongest case combines an image of the original with corroborating identifiers.

A practical decision path

The following sequence works whether you are using a national index, a county transcription, or a printed abstract. It assumes you have at least a name, an approximate date, and a place to work with.

1. Write down exactly what you saw. Record the spelling as it appears, the record type, the date or year, the place, the collection or database, and any volume, page, entry, or image number. Note whether you are looking at an index entry, a transcription, or an image of the original. This is the single most useful habit, because it lets you tell later whether two “different” names came from different layers of the same source.
2. Identify the layer. An index entry is not the record. A transcription is not the record. FamilySearch’s guide recommends going back to the original when a copy may have been misread, checking whether the original has been filmed or digitised and, if necessary, writing to the courthouse, archive, or church that holds it.
3. Obtain the image if you can. Compare the index entry with the image, letter by letter. If the image shows the spelling you expected and the index shows something else, you are looking at a transcription or indexing error, and the original form is the one to use.
4. If the image shows the other spelling, treat it as a recorded variant for that document. Then ask how far it extends. Does it appear once, or repeatedly? Is it confined to one clerk, one parish, one census year?
5. Test plausible readings against the handwriting. Use a table of commonly misread letters as a checklist, not as an answer. The guide’s illustration — that CARTER might be searched under GARTER, EARTER, OARTER, and similar forms — is a list of candidates to try, not a claim about any record.
6. Look for relatives. The guide suggests searching the names of parents, children, siblings, and other relatives in the index. If relatives appear, examine the original record to see whether the person you want is mentioned there but was missed or mis-indexed.
7. Cross-check other identifiers. Dates, places, spouse, children, occupation, and migration patterns should agree before you conclude that two name forms refer to the same person.
8. Log the outcome and keep both forms. A simple variant log — form observed, date, place, source, image seen or not, conclusion — prevents you from repeating the same search and from quietly turning a hypothesis into a fact.

If you cannot obtain the image, the honest conclusion is provisional. You may search under candidate forms and note which ones produced results, but you cannot yet say whether the difference sits in the original or in the copy.

A comparison table for your own notes

The table below is an instructional pattern, not a record. Fill it with your own entries.

Pitfalls and limits

• Assuming every difference is a variant. Some differences are errors in the copy you are reading; others belong to entirely different surnames.
• Assuming every difference is an error. Non-standard spelling was normal, and a form that looks “wrong” to a modern eye may be exactly what the record says.
• Treating a search result as proof. A fuzzy match or a returned hit tells you the database contains something; it does not confirm identity or relationship.
• Merging people on sound alone. Similar-sounding surnames are not evidence of a shared family.
• Reading a variant as a legal name change. A spelling difference in a record is not by itself evidence that anyone changed their name formally.
• Confusing given-name and surname variation. Initials, abbreviations, and middle-name usage affect how a person is indexed, but they are a different problem from surname spelling.

What this method can establish is limited. It can show that a particular document contains a particular spelling, and it can show that an index entry differs from the underlying image. It cannot, on its own, prove that two people are the same person, that a family used one spelling exclusively, or that a name has a particular origin.

The next research step

Pick one spelling difference you have already noticed and take it through steps 1 to 3 above: write down the source and layer, then obtain the image or, if none is available, note the repository to contact. That single comparison resolves more questions than a broad sweep of new searches. From there, a variant log built over several records will show whether the difference is a local habit, a single clerk’s rendering, or an index artefact — and that pattern, not any one entry, is what supports a conclusion.

Sources and editorial status

This article is based on FamilySearch’s wiki guide Name Variations in United States Indexes and Records, which describes the causes of name variation in indexes and records and recommends searching original records and relatives’ names. That guide is written for a United States context; researchers working elsewhere should check the equivalent guidance for their archive or database. No individual historical record is cited here, and the table is a blank template rather than a completed research finding. Editorial review pending.