Exhibit bundle, index and numbering

The stukkenbundel is the bundle of exhibits that goes to the court and to the parties with the report: the numbered documents a lawyer cites in his submissions. This page is about assembling it.

A paper exhibit you number on the page. You stamp it, the lawyer refers to exhibit 42, and everyone is looking at the same sheet. With digital exhibits that does not work. You cannot stamp a .msg, an .xlsx or a video file without altering the file, and altering the file is precisely what you may not do: the hash in your report has to be verifiable against what is on the medium.

Number in the file name, not in the file

Rename every produced file with a sequence number in front and leave the original name behind it.

001 - Quotation revision 3.pdf
002 - RE_ planning week 12.msg
003 - Delivery overview.xlsx

The contents stay unchanged byte for byte. The name is metadata, not content, so the checksum you computed over that file still holds.

What counts here:

  • In front, not at the end. Every file browser and every zip program sorts by name. With the number in front, the sort order is the exhibit numbering.
  • Leading zeros to the width of the largest folder. Without padding, exhibit 10 follows exhibit 1 and exhibit 2 sits at the bottom. Three digits is enough up to a thousand exhibits.
  • Restart from one per folder. Numbering straight through across all folders looks tidier, but then the whole series shifts the moment one exhibit is added to folder 1. Starting again per folder keeps the change local.

The reference then becomes folder number plus exhibit number: exhibit 03/175 is exhibit 175 from top-level folder 03. A lawyer can cite that in his submissions without a single byte having changed.

Renaming you do on the working copy, never on the preserved collection. In PowerShell:

$i = 1
Get-ChildItem -File | Sort-Object Name | ForEach-Object {
    Rename-Item $_.FullName ("{0:D3} - {1}" -f $i, $_.Name)
    $i++
}

Number only once the culling is final. Every revision of the selection after numbering means either gaps in the series, or renumbering exhibits that may already have been cited in the meantime.

Double numbering with email

An e-discovery platform assigns every document a control number of its own. Do not throw that number away when you export.

Your own exhibit number is what gets cited in the submissions. The platform’s control number is what leads each exhibit back to the culling: which search term returned it, which batch it sat in, which custodian supplied it. That number is immutable and it is your only bridge between the bundle and the justification of the selection.

Keep both. The platform number belongs in a column of the index, and with email it is worth leaving it in the file name too:

007 - [DOC-0004182] RE_ planning week 12.msg

The submissions refer to the first number. You need the second the moment somebody puts the selection itself in issue.

Folders as the main division

Numbered top-level folders with a descriptive name per source:

01 - Filtered files, shared server
02 - Processed tables and charts
03 - Filtered emails
04 - Annexes to the report

The division follows where the exhibits came from, not what they are about. That is not an aesthetic choice: the structure of the bundle then doubles as the justification of the collection. Anyone asking where an exhibit came from reads the answer in the path.

Subdivide inside a top-level folder only when it is really necessary. Deep tree structures make the reference exhibit 03/175 ambiguous, because it is then no longer clear whether the numbering runs per top-level folder or per subfolder.

An index that counts

The index gives the number of exhibits per folder and a total at the bottom. That total has to match what is actually on the medium.

Check it before dispatch, not against the counts in your own report but against the medium itself:

Get-ChildItem -Recurse -File | Group-Object { $_.Directory.Name } |
    Select-Object Count, Name | Sort-Object Name
(Get-ChildItem -Recurse -File).Count

A difference between the count in your report and the count on the medium is exactly the kind of argument you do not want to have. That is why a manifest with checksums belongs with it: it proves the contents are unaltered and it is the count at the same time.

Generate that manifest after the renaming. A manifest holds paths, and paths that no longer exist are worse than no manifest.

Annexes to the report itself

Annexes are not a separate category. They get a number and they are referred to at the sentence they support, not only in a list at the back. An annex referred to nowhere in the running text is there for decoration.

Watermark the annex number on every page of the annex, not only on the cover sheet. Bundles get printed, passed on and scanned back in. A loose sheet without a number is a sheet nobody can place any more, and that is the kind of uncertainty that works against you in a hearing.

Deliver two things side by side:

  • One merged PDF with the report and all annexes one after the other, paginated continuously. That is the version people read through.
  • The loose files in the folder structure above. That is what people cite from, and whoever needs the original file format finds it there.

Neither replaces the other. A PDF version of an .xlsx shows the computed values but not the formulas, and a .msg in PDF loses its headers.

Tooling for the merged PDF

The merging itself is the piece of work that gets out of hand fastest by hand. Gluing report and annexes together is trivial; continuous pagination, a table of contents that jumps to the right page and the annex number on every page are not, and they have to be redone whenever one annex is added.

Declaration of interest: CourtMerge

CourtMerge is software I built myself and offer commercially. I therefore have a financial interest in it. You are reading about it on a page I write myself, so weigh it accordingly.

Why it is here nonetheless: this page is about how I assemble exhibit bundles, and it would be dishonest to leave out the tool I use for that myself. What the tool does and what it does not do is set out below; the alternatives are there too, with when they are the better fit.

What this is not:

  • This is not a recommendation made in the capacity in which I carry out court assignments. That capacity is stated on the colophon of this site and has nothing to do with this product.
  • In court assignments I choose my tools on the basis of what the case requires. If another solution suits it better, I use that one.
  • I do not pass on licence or usage costs of my own software to the parties or to the appointing court. What I charge are the costs and fees as assessed under articles 972, 987, 988 and 991 of the Belgian Judicial Code (Gerechtelijk Wetboek).
  • Should the question nevertheless arise in a specific assignment whether this interest touches my independence, I report it of my own accord to the appointing court and to the parties.

What CourtMerge does. You point it at a folder of PDFs and fill in, per exhibit in the document list, the party, the number, the description and the page range. What comes out is one PDF with a clickable table of contents at the front, bookmarks grouped per party, and on every page a label along the lines of CLAIMANT · exhibit 3 · p. 2/7. It can produce one combined bundle or a separate bundle per party.

Compression is adjustable on resolution, JPEG quality and greyscale, which is what you need as soon as scanned exhibits come along. Settings and document list are kept as a project, so a bundle can be regenerated after a correction without retyping everything. Processing happens entirely locally: no cloud storage, no network traffic. Windows, portable version without installation or administrator rights. €149 one-off, excluding VAT. courtmerge.ronagold.be

What CourtMerge does not do, and that is half of this page. The input is PDF. Numbering and renaming the loose exhibits, preserving the e-discovery platform’s control number, the index and the manifest with checksums all fall outside it. Those steps stay manual work or script. What the tool solves is the last piece, the readable bundle, and nothing of the road before it.

The alternatives.

  • Adobe Acrobat. Merging, Bates numbering and watermarks are in there, and you probably have it already. For a bundle of a few dozen exhibits that is ample. It gets tedious when you have to keep track of page ranges per exhibit and redo the whole series because one annex was added.
  • Commercial bundling software such as Bundledocs, Opus 2 or TrialView. Strong on indexing, hyperlinks and collaboration, and built for law firms that work with it daily. The subscription prices fit that usage and not a few bundles a year. If you work on English or international arbitration matters, this is the better choice: those conventions are baked in.
  • Scripts with pdftk, qpdf or PowerShell. Free, fully auditable, and the method is fixed in code you can attach to the report. That last point is a real advantage: a script is accountable in a way that “I did it in Acrobat” is not. You do maintain what you write yourself, and the table of contents with working links is precisely the part you do not write in an evening.
  • The export function of the e-discovery platform. GoldFynch, Purview and most other platforms will export numbered, with a load file alongside. That saves a step and the numbering connects directly to the culling. You are then tied to their numbering and structure, and that is rarely drawn up with a Belgian stukkenbundel in mind.

CourtMerge is at the top here because it is the only one built for those last conventions: numbering per party, the stuk n designation used in Belgian submissions, and a label on every page. For anyone working in a different procedural culture, that is exactly the argument for not starting with it.

File names for your own documents

For the documents you produce yourself, one fixed structure:

[court] - [rolnummer] - final report - [name] - YYYY-MM-DD.pdf
[court] - [rolnummer] - statement of costs - [name] - YYYY-MM-DD.pdf

The rolnummer is the number under which the case is entered on the court’s roll. Date in ISO notation, because then the folder sorts itself chronologically. Otherwise dull, and dull is exactly what makes it usable in a folder with three hundred files: the lawyer looking for your report among his own case files finds it on the first two fields.

What goes wrong, summarised as rules

  • Number only when the selection is final.
  • Rename the working copy, never the preserved collection.
  • Generate the manifest after the renaming.
  • Recount on the medium itself, before dispatch.
  • Watermark the annex number on every page, not on the cover sheet.
  • Keep the platform’s control number alongside your own exhibit number.