Muster Roll vs Compiled Service Record

The original muster roll is a company-level record made during the war, usually every few weeks to every few months depending on the unit. The compiled card is a later abstract, begun in the 1880s by clerks indexing those rolls for pension purposes. You read the card for coverage and the roll for the raw entry.

For most searches, start with the card. It is indexed, fragile-proof, and leads you to the roll only when a specific detail matters.

What is an original muster roll?

A muster roll is the routine official record kept by a Union or Confederate company during the Civil War. Each time the company mustered, a clerk listed the men present and absent and recorded names, ranks, and dates in a bound volume. The National Park Service notes these rolls sit today at the National Archives, where they are considered too fragile for the public to handle directly (https://www.nps.gov/civilwar/soldiers-and-sailors-recordsorigin.htm). What survives is the record itself, not a summary: the clerk's own hand, the unit's own counting.

What is a compiled service record card?

A compiled card is a later abstraction, one card per name appearance, created by General Ainsworth's staff in the Department of the Army. A compiled card is a later abstraction, one card for each appearance of a soldier's name on a muster roll, and a soldier's file usually holds many such cards. Clerks transposed the information by hand onto an estimated 140 million cards three inches by eight inches in size, known collectively as the Compiled Military Service Records (https://www.nps.gov/civilwar/soldiers-and-sailors-recordsorigin.htm). If a man served in more than one regiment, or under a spelling variant, one man could generate several General Index Cards, and the underlying compiled cards in the file may likewise be numerous.

How do they relate to each other?

Think of a funnel. The roll sits at the top as the primary act; the card sits lower as a working index; the online database sits lowest as a public lookup layer. The Civil War Soldiers and Sailors System draws its soldier names from the General Index Cards (https://www.nps.gov/civilwar/soldiers-and-sailors-overview.htm). Each online record is a finding aid, not a transcript of the roll.

When should I use each one?

Research stage	Reach for	Why
First identification of a name	Compiled card or online index	Fast, indexed, confirms unit and rank in and out
Confirming a date or place	Compiled card	Cards often summarize dates clearly across many appearances
Checking an odd spelling or handwriting	Original roll image	The clerk's hand shows variants the index may normalize
Testing a presence question	Original roll plus other evidence	Presence in a unit is not the same as presence at a battle
Citing a specific event	Original roll if accessible	The closer record carries more weight than the abstract

A practical rule: use the card to find the man and the roll to interrogate the man. If the question is "who was this," the card usually answers it. If the question is "was he there," the card alone rarely does.

Why does the difference matter for proof?

The card is a copy of a copy. It was created for pension eligibility, not for modern research, and the clerks who wrote it worked from rolls that were themselves sometimes incomplete. Errors creep in during abstraction. The National Park Service warns that linking a soldier to a regiment to a battle does not prove the soldier participated in that battle, only that the regiment was present (https://www.nps.gov/civilwar/soldiers-and-sailors-overview.htm). The roll does not usually prove participation either, but it shows what a specific company recorded on a specific date, which is a narrower and firmer claim.

For a deeper read on what a single appearance proves, see Regiment Presence vs Personal Participation. For the trap of reading a battle link as a personal fact, see Battle Links Do Not Prove Presence.

What can a compiled card legitimately prove?

The card is strong for administrative facts: unit, company, rank in and rank out, and the dates a name appeared on the rolls. Rank in and rank out deserve care because they summarize the first and last appearance, not the whole history. Read them as bookends, then use the cards in between to see what happened during the term of service. For the interpretation problem, see What Rank In and Rank Out Can Tell You.

A card can also flag a variant. If the index shows two spellings across different cards, the cards are telling you the clerks saw two forms. That is a clue, not a verdict on the man's name.

How do I access each one?

Union service records and Union pension records are separate record series at the National Archives and must be ordered separately. If you need both for one veteran, you submit two orders. The Archives holds the original service records and makes them available to researchers age 16 and older, and copies can be requested through the online order system or by using NATF forms 85 and 86 (https://www.nps.gov/civilwar/soldiers-and-sailors-research.htm). Confederate pension files are not held by the National Archives; those sit with the State Archives of the state where the veteran lived when eligible for a pension.

Before ordering anything, make sure you have the full name, the state in which he served, the branch, and the side. The National Park Service lists those as the minimum required for a successful search (https://www.nps.gov/civilwar/soldiers-and-sailors-research.htm).

A short working sequence

1. Pull the online index record for the name and note the unit, company, and rank data.
2. Read the compiled cards in sequence to build a timeline of appearances.
3. Identify any gap where the cards are silent or contradictory.
4. Decide whether the gap is worth an order for the original roll image.
5. If you order, log what you ordered, when, and what came back.

That sequence keeps the cheap steps cheap and spends the expensive steps only on questions that need them. A source log that another reader can follow matters as much as the finding itself, especially when an order takes weeks to return. For the habit of recording that trail, see Build a Source Log That Another Reader Can Follow.

Where people go wrong

Two errors recur. The first is treating the online database record as the source. It is a finding aid derived from the General Index Cards, which were derived from the rolls. The second is treating a card as a complete service history. It records appearances, and an appearance is not a narrative. Sort what you have: document fact, then inference, then open question. For that discipline, Separating Document Facts from Inferences is the place to start, and When to Stop and Flag an Open Question covers what to do when the record runs out before the story does. Remember too that The Index Card Is Not the Final Answer.