GROUPLAB DOWNLOAD
More

Research › Range tests

The range test log

On this page

GroupLab project. Researched and written with Claude. Testing and data collection by the developer. Written 2026-09-22; the data is from Running log; latest entry 2026-10-04.

No figure: This is a log of what was shot and where it was written up. The figures belong to the articles each entry points to, and are drawn there from the same data.

What we found

four range days and a print test, each with what it was meant to find out and where the answer is written up. On the latest, GroupLab found 137 of 145 holes in photographs of its own sheets taken at the range, with no false mark, and proposed 138 marks for 14 shots on store-bought targets it could fit.

How sure we are

each entry is what was recorded at the time; the counts are GroupLab's own scoring against the shooter's count of shots fired. Where a result was later corrected, the entry says so.

The data

Question: does a GroupLab sheet print at its true size, and can GroupLab detect when it does not?

What was done: test sheets printed at 100 percent and deliberately at 96.2 percent, scanned at 300 and 600 dpi.

What we learned: the developer's printer at "Actual size" measured 1.0001 horizontally and 1.0006 vertically. GroupLab measured the reduced sheet at 0.96200 (600 dpi) and 0.96201 (300 dpi).

Written up in: "Does your printer print at true size?", "Scanner traps".

First range sheet#

Question: does a real, shot sheet read the way generated test sheets do?

What was done: nine shots on a 25-bull letter sheet, scanned flat.

What we learned: this first real sheet started the work on how GroupLab reads holes in real paper, rather than in simulated targets.

Range day, 2026-09-20#

Question: how does GroupLab do across calibers, with shots that land between bulls, with a blank zero sheet, and with phone photos against scans of the same sheets?

What was done: five 25-bull load development sheets (GL-CF25-LTR-D), one blank zero sheet and one commercial target, all at 100 yards. Every sheet was scanned at 600 dpi and photographed with a phone in several bursts through the afternoon, 59 photos in all. A second phone photographed the 6mm Creedmoor sheet.

Sheet Cartridge Load as written Shots What it tests
1 6.5 Creedmoor, 28 in Seekins 153.5 gr LRHT, 42.4 gr H4350, Alpha brass, 7.5 BR primer, 2.874 in 15 One shot per bull, rows 1 to 3
2 6.5 Creedmoor Zero group A blank sheet with a hand-drawn cross
3 6.5 Creedmoor Same as sheet 1 25 A full sheet; now GroupLab's sample scan
4 .22 LR, 20 in CZ SK Standard Plus, 40 gr 23 Strong crosswind, windage changed after row 2
5 6 ARC, 18 in RTR 108 gr ELD-M, 27 gr N140, Starline brass, GM205MAR, 2.250 in 20 A load the rifle was not zeroed for: shots land nearer the wrong bull
6 6mm Creedmoor, AI AXSR 120 gr LRHT, 39.0 gr N550, Lapua brass, 2.810 in 10 Five shots with GM205MAR primers, five with CCI BR-4

What we learned:

  • GroupLab's counts matched the shooter's on every scanned load sheet, after two detection defects found on these sheets were fixed.
  • Assigning each hole to its nearest bull gives confident but wrong groups when shots land between bulls (sheets 4 and 5). GroupLab now lets you say which bulls you aimed at.
  • Switching primers on sheet 6 moved the point of impact clearly, while the difference in spread could not be shown from five shots each.
  • Photos place a sheet almost as precisely as scans, but hole sizes in photos depend on the light.
  • The original bull was hard to see through lower magnification and less expensive glass.

Written up in: "Why a photo cannot tell you your bullet's size", "A bullet hole is not the bullet", "Did the primer matter?", "When shots land on the wrong bull", "Scans against phone photos", "Wind or rifle?", "Zeroing on a blank sheet", "Pooling groups", "How to photograph a target", "Scanner traps", "Can you see the bull?".

Range day and aim point test, 2026-09-26#

Questions: does swapping suppressors move the point of impact, which aim point designs can two shooters center on through four scopes, and how does a real 25-shot sheet fare when every shot lands nearer the wrong bull?

What was done: three 25-bull sheets at 100 yards on an OSB backer, and the aim point card scored by the developer and a friend.

Sheet Cartridge Load as written Shots What it tests
7 6 ARC, 18 in AR-15 105 gr Aeromatch, 24.2 gr N135, Starline brass, GM205MAR, 2.250 in, Thunder Beast Dominus K 25 The first suppressor, shot first
8 6 ARC, 18 in AR-15 The same load, Magnus S 25 The second suppressor, 20 to 30 minutes later
9 6.5 Creedmoor 153.5 gr LRHT, 42.4 gr H4350, Alpha SRP brass, GM205MAR, 2.873 in, Magnus S 25 Every shot about 0.9 in high: nearer the bull above than its own

The 6.5 load is not the one of sheets 1 and 3: those were written with a 7.5 BR primer and 2.874 in, so sheet 9 is not pooled with them.

What we learned:

  • With the Magnus S the 6 ARC shots centered 0.28 in lower than with the Dominus K, p = 0.049, and the spread did not change. The order the sheets were shot in cannot be separated from the suppressor.
  • At 10x the current bull could not be centered through any of the three high power scopes. Designs whose center subtends about 3.4 arcminutes or more could. Through the PLxC at 4x only the 2 inch bull could.
  • A fine crosshair reticle covers a design whose center is a small feature.
  • On sheet 9 GroupLab found 23 of 25 holes and gave 5 of them to the bull above their own.
  • Three sheets printed from one PDF carry one design identifier; the developer told them apart by writing K, M and C in the serial box.

Written up in: "Did the suppressor move the point of impact?", "Can you see the bull?", "Aim points for 1x to high power optics".

Range day, 2026-10-04#

Questions: how well does GroupLab read photographs of its sheets taken at the range, from close and from a little farther away; how does it do on store-bought targets it did not print; and does the chronograph's own export of each string open as it comes off the device?

What was done: three GroupLab sheets, each photographed at about 1.5 and 3 feet: a C bull sheet (15 shots, .300 Norma Magnum, about 2800 fps), an E bull sheet (25 shots, .223 Remington) and a load sheet, GL-CF25-LTR-D (25 shots, 6 ARC, about 2400 fps). Six store-bought targets: five with .223 Remington at about 3000 fps and one with four shots of .300 Norma Magnum. Three chronograph strings, exported from the chronograph's own results.

What we learned:

  • On GroupLab's own sheets, 137 of 145 holes were found in the photographs, with no false mark: the C bull 15 of 15 close and 14 from farther away, the E bull 22 of 25 at both distances, the load sheet 25 of 25 at both. One torn hole on the C bull was missed in every photograph of it.
  • Of the store-bought targets, one was recognized by its printing; two more were the right product but scored under the line GroupLab requires before it trusts a match, and were refused; one of them had its printing spattered by fragments from a bullet striking steel above it. Two are not in GroupLab's library.
  • On the three store-bought targets GroupLab could fit, Find holes (Experimental) proposed 138 marks for 14 shots, most of them on the target spattered by fragments. It invents far more than it finds on a shot-up commercial target, which is why it stays Experimental.
  • The chronograph's export of a single string, with its own header and number format, now opens in GroupLab.

Written up in: not yet as an article; the scores are recorded in the project's results log, entry 374.

Aim point test, planned for 2026-09-23#

Question: which aim point designs can shooters see and center on at 100 yards, across scopes of different magnification and glass?

What was done: a card of nine designs at 100 yards, scored through a Vortex Razor HD Gen III 6-36x56, a DNT TheOne 7-35x56, a Vortex Strike Eagle 5-25x56 and a Primary Arms PLxC 1-8x24, plus a friend's scope, by two observers.

What we learned: it was shot on 2026-09-26 instead; the results are in the section above.

Written up in: "Can you see the bull?", "Aim points for 1x to high power optics".

Next#

  • Aim point tests for 1x red dots and prisms, low power and medium power variables, at distances suited to each.
  • More sheets per load, so loads can be compared with enough shots to mean something (see "How many shots do you need?").

Data#

  • data/range-day-2026-09-20.csv: the six range-day sheets, cartridges, loads as written, shots and notes.
  • The 2026-09-26 sheets: suppressor-shift/data/offsets-2026-09-26.csv, and the aim point scores and groups in can-you-see-the-bull/data/.

What this means#

This page is evidence, not advice. It exists so that the numbers quoted elsewhere on this site can be traced to a day, a rifle and a sheet, rather than appearing as figures from nowhere.

What to take from it: how much goes wrong on an ordinary range day. Sheets that could not be read, a scanner that cropped without saying so, a zero that was not what it was assumed to be. Every one of those became a rule somewhere else in this project, and none of them would have been found by reasoning.

If you are running your own test, the transferable part is writing it down at the bench. Nearly every difficulty here was recoverable only because something was recorded at the time.