Thursday, June 24, 2010

Census Search Tips - Part 1

Case Study 1: John Gillon Owston – 1860

Over the years, I have developed some searching techniques to find missing relatives in the various census records that are hosted on While I have not been 100% successful, I have had the opportunity to find several allied family members by using some different techniques.

Because certain family members were impossible to find in various census records, I had written off these relatives as not being enumerated during the years in question. Anyone who has conducted genealogical research for any length of time will appreciate the frustration of not being able to locate a certain person in a particular census year.


Since I started researching my family in 1978, I had been fortunate to have been able to find my great-great grandfather in every extant US census from 1850 through 1900 – with one major exception – the 1860. There were a couple of variables that I knew. In 1860, he was living in Detroit, as his first wife had died in that location during October of that year. This was documented in both the Pittsburgh and the Detroit newspapers.

Second, he fathered a child with a woman who would become what I thought was his second wife (I later found out in late 2009 that she was his third). This son, who was born in the greater Pittsburgh area in November 1860, would have been conceived in February 1860. It was later determined that this son was not his biological child, but rather was unofficially adopted by my second great grand father. See the post on Pruning the Family Tree with DNA evidence. Therefore, it appeared that John G. Owston left Pittsburgh for Detroit at some point during 1860 [now believed that he went to Detroit via Canada by 1859 and left Pittsburgh in 1857].

With the information known at the time, it was reasonable then to believe that he was missed in the census that year having moved prior to the Allegheny City census enumeration and settling in Detroit after the census was taken in his new locale.

 John G. Owston, circa 1900

To answer this and other family questions, I made a genealogical journey from West Virginia to Michigan to see family and to do research during May 1995. Since online genealogical sites were several years away, the most cost effective way to do a shotgun approach to genealogical research was to bite the bullet and do a road trip and do the work myself. The Detroit Public Library, one of my stops, had a wealth of information: city directories, necrology files, and an 1860 census index.

I searched for John in the census index – I was careful to use all of the possible variations and misspellings of our surname (nearly 30 in all). Since this was a paper index of the heads of households, the search didn’t take too long as the variants were often side by side.

Zilch, nada, nothing.

Although I had found him in several city directories and located the obituary for my great-great grandmother, the search in the census was futile. The trip was not a total loss and I just chalked it up to my supposition that he must have been missed in the census.


The 1860 census was critical as I had supposed that my 2nd great-grandparents had at least one other child – a daughter named Frances or Fannie – a name that was used often in the family. John’s mother was named Frances and John had four nieces also named Frances. All were nicknamed Fannie or Fanny.

My supposition of John having a daughter named Fanny was based on a hair sample glued to a card stock found within the pages of an old family bible. The note attached to the hair read, "Fanny Owston 1859." I was curious to find if this was indeed John's daughter or if it was one of his nieces. Not finding her in the census would not prove she wasn't his daughter as she could have died between 1857 and 1860; however, finding her would prove that she was.

Also during my trip to Detroit, I found a necrology listing for child listed as Benjamin F. Oustin. This was new information and I was curious to know if another son might be found in the census records or if this was a spurious listing of an unrelated person with a similar surname. Further research on Benjamin indicates that his surname was Austin and was unrelated (as far as we know) to our family.


By 2006, I had worked through some census searching techniques in order to find 1,200 Civil War soldiers of whom I was tracing from birth to death. I decided to use these techniques in order to see if I could find John Gillon Owston and his family in 1860.

Having been through the Pittsburgh records with a fine tooth comb years earlier, I was convinced that, if he was anywhere in the census records, he had to be living in Detroit. To find John, it required deduction (that he must be in Detroit) and induction – using what I knew about him and his family.

The family members, of which I was sure, included the following:

  • John – born 1826 in Canada; a railroad employee
  • Martha – born circa 1831 in Pennsylvania
  • Newton also called by his middle name French – born 1854 in Pennsylvania


Since the various city directories listed John as living within the city limits of Detroit and not a suburb, I concentrated on finding him as a resident of the city. I had tried the Ancestry search by looking for John Owston under all of the variations of the surnames using the wildcard (*).

This is a nice feature for searching any of Ancestry’s records; however, you need at least the first three letters of the name (i.e., Ows*). This search brings in every surname in that begins with these three letters. You can also limit your search by using the wildcard within the middle of the name – such as Owst*n – which narrows the search considerably. The caveat is that the name must at least begin with three letters to do a wildcard search.

I still couldn’t find him. This included using the variations of Aust*n. I even tried a search on only the surname and its variations, but I was still was unable to find his family.

During my Civil War research, I discovered that searching through a municipality or county by only a first name proved fruitful in finding an individual in lower populated areas. The more obscure the first name, the better. When using this technique, I often searched for a range of years of birth and nativity to narrow the search.

The problems with this particular search was that Detroit was not sparsely populated and John's first name was fairly common. Even when narrowing the search to a 10 year range set at 1826 to plus and minus 10 years and adding his birthplace of Canada, all searches proved fruitless. I made the 10 year search as the 1870 census listed him as 10 years younger than he actually was.

In 1860, there were 2881 Johns in Detroit; 1008 were born between 1816 and 1836; and 49 born during this time were from Canada - none matched. I also searched under his initials (J.G. and J.) without any luck. Since that time, Ancestry's search algorithms have been adjusted and I will discuss this below.

For my Civil War project, when I was not able to find a soldier under his own name, I often searched under his wife’s name. By doing this, I was able to find a number of men who were listed under their middle names or initials. The problem with this particular search was Martha was a common first name of the period and this was a metropolitan area – the search might take forever. I did this but narrowed it to her birth year of 1831 +/- 2 years. The return brought up one unrelated Martha.

I decided to search for his son. Like searching for his spouse, searching for a child was an option that could not be conducted via paper indices of census household heads. My great-grandfather's name, while not unusual, was not very common. He was named after his mother’s family. Newton was his great grandmother’s maiden name and mother’s middle name and French was his mother’s maiden name. He went by both and is enumerated by both names and his initials in subsequent census records.

I framed my search by using Ancestry’s advanced search. Besides the geographic location, the search parameters included the following: the first name of Newton, the year of birth as 1854 with a plus or minus of 2 years, and his nativity of Pennsylvania. I also made sure that my search was narrowed to the exact spelling of his first name by checking "Exact matches only."

The search produced nothing, so I tried my great-grandfather’s middle name of French with the same parameters.


The search was successful as only one name was returned.

  • Name: French Austin;
  • Home in 1860: Ward 4, Detroit, Wayne, MI;
  • Age in 1860: 6;
  • Birth Year: About 1854;
  • Birthplace: Pennsylvania;
As my surname is commonly misspelled as Austin, this was the lost census record that I had been earnestly seeking. Previously, I had been unable to find John. The reason became clear as his initials were reversed as "G.J." in this record. (Under the search conditions in 2006, a search of J.G. would not return G.J.; however, the latest search algorithms automatically returns reversed initials as well as in the requested search order). The data confirmed that it was him – the subject was aged 35, employed in railroading, and was a Canadian native. In addition, the family triangulated the results as being correct.

  • Martha Austin, age 26* [sic]
  • Fanny Austin, age 8
  • French Austin, age 6
  • Katherine McEnaly** [sic], age 57
*See my post on Census Basics explaining why ages are often wrong or they vary from census to census. Martha couldn't be found in the search as her age was five years off.

**Katherine McAnulty was Martha’s aunt and was enumerated with her in the 1850 census as well. She was also listed as part owner, along with John Gillon Owston, of the cemetery plot at Elmwood Cemetery in Detroit where my great-great grandmother is buried. 


For 28 years, I had been searching for my second great-grandfather. This find also proved my suspicions that Frances or Fanny, whose lock of hair I have, was John and Martha’s daughter as well. It also verified that she was the oldest child of John and Martha.

Happy searching.

For more on John Gillon Owston, see my listing for him.

Tuesday, June 8, 2010

The Soundex Code: Boon, Bane, or Bomb?

Those of us who had to do research the old fashioned way on clunky microfilm machines learned that the later U.S. census records were indexed with a coding system called the Soundex. The 1880 census was the first to be coded by the Commerce Department with Soundex cards. Unfortunately, the 1880's indexing was limited, as only households with children 10 and under were cataloged by Soundex numbers. If you were searching for a household that had older children or no children, you were out of luck with regards to the 1880 census Soundex.

Unlike the 1880, the censuses for 1900, 1910, 1920, and 1930 were completely Soundexed. These cards were later microfilmed by the National Archives and a person could search the records first by state, then by the Soundex code, and finally by first name and middle initial of the individual in question. If you were doing research on a specific surname, you could look at the Soundex microfilm and get a truncated view of the census records for everyone by that surname by that state.  Additionally, the cards provided locality data and page numbers so a researcher could reference the actual census record for more complete information.

The Workings of the Soundex

The Soundex system was developed in the second decade of the 20th century as a phonetic algorithm that combined sounds with an index tool. A Soundex number is constructed with the name’s first letter and a three digit numerical code based on consonants. Vowels and the consonants H, W, and Y are not coded. Double consonants or two consonants of the same Soundex code are counted as one letter. For example, “TT” would be coded as one letter and “CK,” in which both letters make up the same code would be counted once. If the word has more than three coded consonants, the remaining consonants are ignored. The numeric codes are based on the following:

Letters  Number  
  B, F, P, V       1
  C, G, J, K, Q, S, X, Z          2
  D, T       3
  L       4
  M, N       5
  R       6

For example, the following names are coded as such:

Name  Soundex  
   Archibald         A621
   Boston         B235
   Carothers         C632
   Dickenson         D252
   Schmidt         S253
   Milosovic       M421
   Wineberg       W512

When names do not have enough characters to fill three Soundex code numbers, the blank space is represented by a zero.

  Name    Soundex 
    Bates         B320
    Poe        P000
    Goins         G520

A Major Genealogical Help in the Analog Age

When I immersed myself in the world of genealogy in 1978, only the 1880 and 1900 census records were available. The Soundex for these two censuses allowed me to gather data quickly on my family. When the 1910 became available in 1982, I visited libraries and viewed these records. The same occurred with 1920, which came available in 1992. With this census, I ordered the microfilm records directly from the National Archives through inter-library loan.

In 2002 when the 1930 census was released to the public, and both offered this particular census online. Since I was subscribed to both, it was no longer necessary for me to sit at a microfilm reader to use a census Soundex. actually beat Ancestry to the punch and had a limited version of the 1930 census online first.

If I remember correctly, their images were better than Ancestry's. Ancestry returned this favor by buying this competitor in 2003. The sites were run as separate entities for about two years until the records of both were consolidated.

Is the Soundex Viable in the Digital Age?

Theoretically, with complete indexing of census records, the Soundex is almost obsolete, but not entirely. The Soundex system is used on all of the records on and may be helpful if you are searching for a name that is often spelled differently but is phonetically the same. While the coding is not necessary for us to know these days, it is helpful to understand how it works and its limitations. 

For example, my surname Owston and its variant spelling of Ouston is coded in the Soundex as O235 – but so is Oston, Oyston, Ogden, Oxton, Osteen, Ostani, etc. Some of these are close to my surname and others are not. Sometimes my name is misspelled as Auston and the code A235 will return Austin, Austen, Auston, Academy, Acton, and many, many others.

The upside of Soundex searches is that it returns names that are actually similar to the name in question. The downside of Soundex is that it returns names that are phonetically equivalent but dissimilar to the name in question – such as same code of O235 representing both Ogden and my surname of Owston.

Using the Soundex option in searches on Ancestry with other specific information may help you find a long lost relative whose name may have been misspelled in the census. Ancestry's Soundex option allows multiple types of records to be searched and not just census records. It also gives the researcher the opportunity to simultaneously search census records beyond the borders of one state - something that was impossible on microfilm as each state had to be searched individually. The key to using this feature successfully is to provide enough other information to narrow your search.

How to narrow that search and a few other secrets to sleuthing the census records is forthcoming in our next census installment.