Ancestor Search:
Enter as much information as you know about your ancestor and click search:

  First Name: Last Name: Location:   

Buy the Red Book from and save $8 off the price!
From the Ancestry Daily News
  Michael John Neill - 12/22/2004

Paging through Spelling Variations

Ulfert Behrens and nineteen of his relatives are enumerated on the same page of the 1860 census, spread over four households. The last name is not spelled the same way in any of the households--and never correctly. This week we take a look at these entries and discuss how lessons learned from these families might help others searching for their own missing census entries.

The extended Behrens family is listed in four separate households on page 884 in Clayton Township, Adams County, Illinois in 1860:

Gerd Behrens (the father) is enumerated as Gerrett Barnes.
Herman Behrens (a son) is enumerated as Herman Barams.
Ekke Behrens (a son) is enumerated as Ekke Barse.
Ulfert Behrens (a son) is enumerated as Woolpert Barrus.

While different readers may interpret these renderings in different ways, there is no doubt that the last name is never written as Behrens. Readers who want to view the census entries and see them for themselves can do so here on my website.

Why the Difference?
In this case, the variations were most likely due to phonetic variations of the last name. The individuals involved most likely pronounced their last name Behrens in a way that sounded like “barns” or “bar ands.” Given these pronunciations, the renderings of the last name (Barnes and Barams) for the father and first son are reasonable considering the census enumerator likely was unfamiliar with the language. The two additional spellings for the last name (Barse and Barrus) appear reasonable as well, but would not have been high on my list of potential variants.

Transcription Problems
In addition to phonetic various, it's important to keep in mind that some letters are easily misread, which can cause indexing problems. The following letters are among those easily misread, but also remember that individual handwriting can significantly influence the way a name is interpreted by an indexer.
- Upper case letters
F and H
J and I
J, G, and Y
K and R
S and L
O and Q
P and R
T and F
U and V
W and M, UU

- Lower case letters
a and ee
a, o, and u
b and f
d and el
j and i
k and t
s and l
t and c
ss, fs, ps
w and vv
y and g

Keep in mind that if the handwriting is really bad, some letters can look like just about anything. A manual search of the actual records is sometimes necessary when there is good reason to believe the individual should be listed in the original record but is not listed in the index.

What Kinds of Searches Would Have Found These People?
I initially located these people the old fashioned way--I searched the census microfilm page by page (or image by image) for those four townships in Adams County, Illinois, where I thought the family should be living in 1860. When I saw any non-English name I said it out loud (albeit softly) to determine if the last name was a potential match. If I had used online indexes my approach would have been different.

What Kind of Online Approach Would Work?
With an online search, obviously searching for Behrens as an exact match would not be a successful strategy. A search for Behrens using a Soundex option would have located the entry for Barnes and Barams. In reality, once these two entries had been located the others would have been seen as well. (Remember: A genealogist always looks on the entire page and on a few pages before and after the located entry to determine if any family members are living nearby.) Locating Ekke Barse and Woolpert Barrus in the online index would have been more difficult.

Would Soundex Have Worked?
A Soundex search for Behrens would not have found Ekke Barse and Woolpert Barrus. Barnes and Barams (and Behrens) all have Soundex code B652. Barse and Barrus have a Soundex code of B620.

How would I have found the Barse and Barrus entries? It depends on where the family lived and how precisely their residence was known.

With rural families, when the location is relatively known, a manual search of the census is the best approach. If the likely residence had not been known with any precision, or the family had lived in a urban area, a broader search might have been in order. In these cases a manual reading of the entire census might not have been practical.

Since the 1860 Census index at is an every-name index, searching for other members in the household is another option. In this case, it would have been somewhat difficult with the household of Ulfert, given the actual renderings of the first names:

Ulfert was Woolpert
Fredericka was Ricke
Trientje was Frartia
Herman was Harman
Gerd was Garrett
Harm was Harm
Volke was Tolke
Claus was Claus
Ekke was Ekke

Unfortunately, Soundex searches are not possible on first names. I also keep lists of variants for first names as well as last name (using phonetic and transcription variants as well). There are several household members here that can be the focus of our search if efforts to find the father are unsuccessful. When the names are non-English names, I tend to focus on these household members first:
- Those whose names are very close to English sounding--Harm is a good example.
- Those whose names have a standard Anglicization. While there were no Johanns in this family, Johann is frequently rendered as John. A census taker may decide for himself how to Anglicize a name, which may not be the Anglicization the family or individual chose.

For those families whose names are English, I tend to focus on the following household members first:
- Those whose names are less likely to have nicknames (shorter names come to mind here), or those whose names have typical nicknames (Sally for Sarah, etc.). Keep in mind that sometimes people are enumerated with a middle name and sometimes these middle names are unknown.
- Those children who are younger at the time of the enumeration. Some search interfaces allow users to enter a specific age at the time of enumeration. A census taker is less likely to be off on the age of a two- or three-year-old child than might be with a fifty-year-old adult.

Do I Know It Is Them?
Funky spellings and “incorrect” information should always cause genealogists to wonder if they have found the right family. There is little doubt in my mind that I have the right families in this case. The age of Gerrett and his wife Twuga (actually Trientje), are reasonably close to what is known from other records. It was also known that they had an adult son Claus who was not married by 1860. A comparison of the household members of the other households confirmed I had the right families (or else there is an extremely huge coincidence). In summary, my conclusion that I had the right families centered on:
- Names and ages in the census that were relatively consistent with information on the families of Gerd, Herman, Ekke, and Ulfert obtained from other sources
- Places of birth were consistent with known information from other sources, particularly which children were born in the United States and which children were born in Germany. These birthplaces dovetailed nicely with the dates of immigration for these families.
- The residence of all four families was consistent with land records and family tradition.

I did not just “grab” some families and make them fit my scenario. I compared the information from the census and analyzed it to determine if it was consistent with information I had already obtained.

Overcoming the Problem
To help overcome problems with spelling variations, keep a list of spelling variants for each last name and first name being researched. These variants should include phonetic variants, handwriting variants, and transcription variants.

Consider manually reading the census if the region is small enough that this is practical or the location is known.

Consider searching for other family members living in the area. They might have been enumerated in the census in a way that is easier to find than your ancestor was.

One Last Note
The household of Woolpert Barrus is continued on the next census page. The first name looks dangerously close to Frautia Barrvis--yet another variant!


Michael John Neill is the Course I Coordinator at the Genealogical Institute of Mid America (GIMA) held annually in Springfield, Illinois, and is also on the faculty of Carl Sandburg College in Galesburg, Illinois. Michael is the Web columnist for the FGS FORUM and is on the editorial board of the Illinois State Genealogical Society Quarterly. He conducts seminars and lectures on a wide variety of genealogical and computer topics and contributes to several genealogical publications, including Ancestry Magazine and Genealogical Computing. You can e-mail him at or visit his website at, but he regrets that he is unable to assist with personal research.

Copyright 2004, Used by the author on his website with permission. Other articles by Michael John Neill can be found here.