1. Linguistic Variations and Transliteration Challenges

• Although there are sources that documented names and place names in English, the dataset contained data that were originally documented in Tamil or Sinhala and translated into English either during documentation or during extraction. This process may introduce:
  • Inconsistencies in spelling and pronunciation: multiple English spellings of the same name due to different transliteration styles (e.g., Sinnathambi vs. Chinnathambi).
  • Loss of phonetic accuracy, especially due to the inability of English letters to accurately capture certain Tamil or Sinhala sounds (e.g., the Tamil name Thambiah may be pronounced incorrectly in English as "Tham-bye-ah" instead of "Tham-bee-ya").
• Where available, the spelling from the original document was preserved to retain reliability, even if alternate standardized spellings exist.
• Transliteration tools and human interpretation both played a role, leading to some unavoidable inconsistencies.
• In some cases, regional dialects or alternate local pronunciations may have influenced how names were recorded or remembered during oral testimony. These variations have been preserved where possible.

2. Spelling and Name Variants

• Some names in this dataset do not clearly indicate which part is the first name and which is the surname. This is because:
  • Naming conventions vary across communities, with some using patronymics (father's name first), while others may use given names first.
  • Tamil, Sinhala, and Muslim naming practices differ significantly and are often influenced by regional or provincial customs.
  • Some documents may list names in full without separators or reverse the order depending on the writer or interviewer.
• As a result, it is difficult to consistently determine the intended order of names in several records. In such cases, the original name order as found in the source document has been preserved.
• Some individuals were found to have more than two known name variants across sources. These have been retained to assist with recognition and reduce false duplicates.
• Users are advised that:
  • Searches by name may return partial or reversed matches, and
  • Apparent inconsistencies in name order do not necessarily indicate different individuals.
• Fuzzy matching was applied to reduce duplication, but manual review is still required when verifying identities across documents with similar names.

3. Date Variations and Ambiguity

• Some records were found to lack precise calendar dates and were described only in general timeframes (e.g., by month, year, or reference period). These entries were retained where reasonably interpretable, but vague or unqualified entries were excluded to maintain overall data integrity.
• Inconsistencies were also noted across sources when matching records by victim name, place of incident, etc., but with date discrepancies due to the reversal of day and month formats (e.g., 7/3/1990 vs. 3/7/1990).
• Where clear uncertainty or vagueness was evident — such as cases marked only as "date unknown", "place unknown", or "victim name unknown" without further qualifying details — such entries were carefully reviewed and, where appropriate, excluded to uphold overall clarity and reliability.
• Partial dates such as month and year (e.g., "June 1989") or just year (e.g., "1989") were preserved where context allowed. When the exact date was missing but could be reasonably inferred, it was noted accordingly.
• Additionally, in some cases, the reporting of a victim's death or disappearance occurred long after the actual incident. This often happened when family members or informants were displaced and only reported the event upon returning to the area. In such cases, the reported date may reflect the range rather than the actual date of incident.

4. Place Name Standardization

• Incident locations were reported using village names, local landmarks, or districts in Tamil/Sinhala/English. These were mapped to official Sri Lankan districts where possible.
• As the data spans events from the late 1980s, it is important to note that administrative boundaries may have changed since the time of documentation. For example, in Batticaloa District, certain GN divisions or border villages that were historically part of the district may have later been reclassified under new or adjacent divisions, or reassigned to Ampara District due to administrative reorganization or resettlement initiatives.
• Some GN divisions in areas such as Valaichenai (Koralaipattu region) were later subdivided or merged to create new administrative units post-1990, often in response to population shifts and development planning.
• In some cases, original Tamil or Sinhala place names may have been altered or replaced during post-war administrative or settlement programs, a process sometimes referred to as name standardization or Sinhalization. Where possible, the original name as recorded in the document has been retained.
• Users are advised to interpret place names with caution, especially when dealing with historical records, as new administrative boundaries and naming conventions may not always reflect the jurisdictional or cultural context that existed at the time of the incident.

5. Data Source Diversity, Verification, and Merging Cautions

• This master list combines entries from multiple sources, including institutional reports, community-led documentation, firsthand victim and witness narratives, as well as secondary reports.
• Source formats included typed reports, handwritten field logs, scanned newspaper clippings, online materials, and spreadsheet-based registries. OCR limitations or scanning errors may have affected certain entries.
• The dataset includes both verified and unverified information. While many entries are drawn from credible and well-documented materials, others are based on unverified narratives obtained from publicly available websites, newspaper archives, or secondhand reports.
• In cases of conflicting details, no single source was prioritized unless corroborated with stronger metadata.
• Their inclusion reflects an effort to preserve all available information, allowing for further verification by victims' families, researchers, and communities.
• Duplicates were reduced using fuzzy matching techniques. However, inconsistencies in names, dates, or place of incident may still result in fragmented or duplicate records. Manual review remains essential in such cases.
• This list does not identify all the dead and disappeared during the period, but only those reported with a complete name and a date and location that fall within the defined time and geography. Records lacking full identification (e.g., "a man," "a child") were excluded. Some victims were listed only in relation to another (e.g., "wife of X," "daughter of Y"), which introduces the risk of duplication when the same individual is named differently across sources. Care was taken to honor all victims while minimizing the risk of overcounting.

6. Ongoing Updates and Community Contributions

• This dataset is part of a continuing effort to preserve, document, and make accessible records of killings and disappearances. Updates are carried out on a rolling basis as new or previously unprocessed data becomes available. This includes both the review of existing sources and the integration of additional materials shared by families, community members, and other individuals willing to contribute further evidence, testimonies, or documentation.
• Users should note that the dataset is not static and may evolve over time as verification improves and new information emerges.

7. Language Availability

This contribution page is currently available only in English, but supporting documents can be uploaded in any language. Tamil and Sinhala translations of the site will be introduced as part of our ongoing upgrade process.