Skip to content
ARMAG

Search matches names, aliases and the text of every page. An arm found by its service designation is found by its alias, not its title.

Editorial policy

These are permanent exclusions, not things awaiting a later version. Each one is enforced by the build where it can be, rather than by anyone remembering it.

Not authored, not linked, not summarised

  • Manufacturing or fabrication instructions of any kind.
  • Conversion to automatic fire, or any modification defeating a regulatory control.
  • 3D-printable files or their sources.
  • Acquisition workarounds.
  • Any content whose purpose is to help someone obtain or modify a firearm outside the ordinary lawful process.

Handloading data falls under the first line: there is no powder-charge field in the database, and there will not be one. Published velocity for a given barrel length is a measurement; a charge weight is a recipe.

No damage, lethality or stopping-power metric. Ever.

Muzzle energy and retained energy are physics, and they appear prominently. Momentum and sectional density are physics too. Stopping power is not.It is a discredited idea with a long history of unsourceable numbers, and any index built on it — however scientific its formula looks — is a claim about wounding dressed as arithmetic.

This includes indices that are conventionally published elsewhere. Taylor's Knock-Out factor is one: it was devised to rank dangerous-game rifles by their stunning effect, its dimensions are momentum times diameter, and it means nothing physically. It is not implemented here, and an automated test fails the build if it ever appears.

No winners

Comparison tables carry no "winner" badges and the matchmaker ranks nothing on effectiveness. Lighter is better for carry and worse for recoil; higher capacity is better for some readers and irrelevant to others. The site does not know what you want, and pretending otherwise is exactly the editorial failure it exists to avoid.

Where a bar is drawn next to a number, its length is that value's percentile within a stated comparison set, and the label is the real value with its real unit. Never a normalised score out of a hundred.

Imagery

Real photographs only. No AI-generated firearm imagery, ever.Image models get variant-level details wrong, which is the exact thing this site exists to get right, and a subtly incorrect photorealistic image does more damage to a reference work than a missing one.

Every image is self-hosted and carries its licence, author and source in the credit panel on the image itself, not in a list somewhere else.