Jewish genealogists who trace family to the early 19th century frequently encounter difficulty trying to follow the trail back to the time when their ancestors did not use hereditary family names. Researchers who find records without surnames often cannot determine if they have found a family member or someone unrelated who has the same given name. Without a surname, how can a genealogist know which Moishe belongs to his or her family? This article describes a project designed to infer and assign family names to records that lack them. In due time, results will be posted on the Jewish-Records Indexing-Poland (JRI-Poland) website (www.jri-p.org). |
I now can trace back 12 generations to the early 1700s/late 1600s with a tree that has more than 900 people. |
In much of Russia, surnames were mandated in official records beginning in 1809, but full compliance was not achieved for many years. Individuals listed in Jewish vital statistics records from Russian Poland during the first quarter of the 19th century (and often much later as well) usually do not have family names. People typically are recorded in birth, death and marriage records only with given names and patronymics (father’s and sometimes mother’s names) only. Occasionally the parents’ patronymics also appear, providing three generations of names, but this practice seems to have depended on a particular clerk’s practices rather than any consistent rule.
I have been tracing maternal ancestors from the towns of Przasnysz (Prushnitz in Yiddish); Chorzele; and three small villages, Dziergowo, Krzynolowga Wielke and Krzyno-lowga Mala, located close to the Prussian border in the area north of Warsaw. Microfilms available at many Family History Centers of the Church of Jesus Christ of Latter-day Saints (Mormons) have been indexed by JRI-Poland volunteers and made available on the Internet. The many records without surnames, however, are not searchable online. When I began the project in 2003, indexed records for some of the towns were available online at JRI-Poland, but not all years were indexed. My first step was to put all relevant unindexed microfilms on permanent loan at my local Family History Center.
Research Goals
A search for Zylberman relatives from the Przasnysz area provided the impetus for the project. I now can trace back 12 generations to the early 1700s/late 1600s with a tree that has more than 900 people. Key was the 1827 death certificate for my great-great-great-great-grandfather Jciek Nosen Zylberman, son of Nosen ben Aron and Basia bat Haim. On the certificate was the Polish-language phrase po sobie, (survived by). Listed as survivors were Jciek’s three sons, Noszek in Krzynolowga Mala, Aron and Wigdor (my great-great-great-grandfather) in Przasnysz; and two daughters, Sura Jzraelowa (Sura, wife of Jzrael) in Przasnysz and Haia Moshowa (Haia, wife of Mosze) in Polny Mlyn. Jciek’s wife, Cywa, was not mentioned; presumably she had predeceased him. With this list of survivors, I connected virtually all the available 19th-century Zylberman records to the three brothers, Noszek, Aron and Wigdor.
The two daughters posed a problem, however, as I could find no hint of their married surnames. Fortunately, they had stayed in the area, so there was hope of finding them. Initially, I assumed that the missing records would be found in the unindexed records, so I began to index the years not yet available on JRI-Poland. The daughters were nowhere to be found.
My next thought was that birth or death records from later years might include the names of the Zylberman females’ parents. For example, if either Haia or Sura’s daughters had children in Prazansyz, the birth records might yield the needed link if the mother’s parents’ names were given. This required the extraction of data from the original rec-ords, because most civil records did not include such information in the indexes, even when the data was included in the original record. Thus began a record-by-record examination to extract the parents’ names found in microfilmed records for these towns. Extraction of parents’ names often revealed the value of extracting the names and patronymics of spouses as well. Data entered into an Excel spreadsheet quickly revealed patterns.
Method 1: Simple Sorting and Inferring Surnames
After entering all the data into a spreadsheet, I sorted the information alphabetically in various ways. A typical sort was by father’s name, mother’s name, then by father’s patronymic; another was to sort first by mother’s name, then father’s name, and then mother’s patronymic. In each (and other sorting arrangements), certain patterns easily emerge that allow one to infer surnames for records that do not include them. Multiple sorting orders are essential because some sort patterns show family relationships that other procedures do not. The JRI-Poland database rules require inferred surnames to be noted explicitly so that users of the database understand that the inferred surnames are not actually listed in the original record. I recorded inferred surnames with the suffix inf (e.g., Zylberman[inf]).
Figure 1 shows a version of the Przasnysz spreadsheet where the file has been sorted according to father’s name, then father’s patronymic. This sort revealed previously unknown children of Leizor Zelkowicz Allensztejn. Note the many alternate spellings for Halka. They required development of additional, new search strategies.
For towns the size of Przasnysz, one may infer a surname reliably if all four given names (father, mother, father’s father, mother’s father) match—so long as only one surname is associated with that combination of names. Where only three of the four names exist in a record, a surname may be inferred reliably if only one surname is associated with that combination of three given names. Inferring surnames based on only two given names is possible in some situations, but not all. I have not ascribed surnames to any records on the basis of a two-name match only.
Numerous surnames may be inferred reliably by simple sorting. More sophisticated strategies must be developed to account for the routinely inconsistent 19th-century spellings, and sometimes clerks made errors. I have found a few records in which Zylberman relatives were listed incorrectly as Zylberstein. I knew the records were incorrect, because all other corroborating data indicated they were indeed Zylbermans.
The phonetic approach to transliterating names generated a wild proliferation of potential spellings. The same individual might appear as Chersz or Ersz or Hersz in different records, even in the same town and written by the same clerk. Another problem arose with double names. Someone might be recorded, at various times as Jciek Dawid, Dawid Jciek, Jciek or Dawid. To handle this problem, a more sophisticated sorting technique can be used in Excel.
Method 2: Auto-Filter
The Excel auto-filter feature (listed under the “Data” menu) helps identify same names with alternate spellings. Numerous sorting orders exist. One may choose to sort, as in the previous method, by father’s name, then by father’s patronymic, then by mother’s name. To capture all the variants of Leizor and group them together, select “custom” for the filter option in the father’s name column and set it equal to L*z*r . Simultaneously, set the filter for the father’s patronymic to Z* where the wildcard * represents any number of characters. This produces all the names (shown in Figure 2) where more of Lejzor Zelkowicz Allenstejn’s children are found. Particularly interesting here, because her marriage was not recorded in Przasnsyz, is the daughter, Golda Fejga, who had apparently married a Curkiel. Note also the numerous variants of Halka that are used for Leizor’s wife, as well as Lejb and Lewin as two variants for Halka’s father’s name.
In addition to the wildcard * that represents any number of characters, the wildcard ? representing just one character, also is useful. For example, filtering on ‘*a?a*’ will capture not only variations such as Chaja, Chaia, Haia and Haja, but also Chaja Sura, Sura Chaia and Ryfka Haja.
Potential for Error
At times, sorting may produce misleading results. A simple sort, or sorting plus auto-filter, involving the death record of a married woman produces her maiden name as an inferred surname, not her married name, because she will be matched with her parents. For example, note in Figure 2 that Golda Fejga (nee Allenstejn) Curkiel had been recorded without the surname Curkiel. Because she was married, the inferred surname goes with her father, Leyzor, and is not assumed to be her married name.
Results
One of my most important inferred records is the 1817 Przasnysz death record for a Zelek Abrahamowicz who died at age 80. A witness at the recording of his death was his 40-year-old son, Leizor Zelkowicz, no surnames given. Other Przasnysz records showed clearly that Leizer Zelkowicz was an Allenstejn. Because Leizer Zelkowicz Allenstejn was my great-great-great-great-grandmother’s brother, I was able to add my great-great-great-great-great-great-grandfather Abraham (father of Zelek Abrahamowicz) as the eldest identified patriarch of this branch of the family, dating back to the early 18th century, if not earlier.
The search for my great-great-great-grandfather Wigdor’s sisters also was successful, if not completely unambiguous. I searched death records for a Sura (or Haia), daughter of Jciek Nosenowicz and husband of Jzrael (or Moszek) by setting autofilter for father’s name equals J* (to capture variants of Jciek) and father’s patronymic equals N*s* (to capture variants of Noszen). Only three Przasnysz records in 58 years exist with a father named Jciek Nosenowicz. (Figure 3). One is the marriage of my great-great-great-grandfather, Wigdor Zylberman, but the other two are the deaths of a Sura and a Haia, both of them married to spouses with the correct name. Both women also are in the right age range. Based on this and other factors, such as children’s naming patterns, I believe that I likely have found the two previously missing daughters of my great-great-great-great-grandfather, Jciek Zylberman.
Martin Meyers of Montclair, New Jersey, has published numerous papers in engineering journals, but this is his first on genealogy. He began doing genealogical research in 2000 and has identified 15 of his 16 great-great-