Individual agent reports from Mississippi sub-assistant commissioners, extracted from NARA microfilm M826 transcriptions (Aug 1865 – Dec 1868)
Sub-assistant commissioners across Mississippi submitted monthly narrative "Reports of Operations" to headquarters in Vicksburg, answering 10 standard questions about conditions in their subdistricts. These reports — the analog to Indian Agent annual reports in the ARCIA — reveal individual agent attitudes toward freedpeople, labor, education, and justice.
The Smithsonian Transcription Center has fully transcribed all four narrative report rolls (30–33) from NARA microfilm M826: Records of the Assistant Commissioner for the State of Mississippi. Volunteer transcribers converted handwritten reports to searchable text, which we parsed from bulk PDF downloads.
| Roll | Period | Reports | Usable | Words |
|---|
Coverage peaks in Oct 1867 when a new circular mandated standardized reports. Reports thin out in early 1865–66 (fewer field offices active) and late 1868 (Bureau winding down).
The "Narrative Reports" rolls contain a mix of document types. Not everything is an actual narrative report — complaint logs, cover sheets, and fragments are interleaved. We classify each report to isolate the material usable for ideology measurement.
Analysis-ready: reports are operations reports or narrative letters with an identified agent and 200+ words of substantive content. These come from unique agents, of whom have 3+ reports.
Agents were identified by cross-referencing report location and date against the NARA personnel roster for Mississippi field offices (148 staff entries, 92 unique names, 45 cities). Signature extraction from report text provides additional identifications.
| Agent | Role | Location(s) | Usable Reports | Avg Words |
|---|
Agent tone varies dramatically. Some agents advocate for freedpeople's rights, document their desire for improvement, and critique white hostility and corrupt courts. Others adopt a paternalistic frame — emphasizing freedpeople's "ignorance," characterizing labor problems as idleness, and echoing planter grievances about dependency. These differences are what LLM-based ideology scoring will capture, and what we expect to predict variation in contract outcomes.