WIND DRIFTThe same arrow was getting two different answers

• Build an arrow in Guided Build, build the same arrow in The Lab, get different wind-drift numbers. That is what was happening, and a customer caught it. The physics was never wrong — the two screens were each doing their own arithmetic to feed it, and one of them was feeding it the wrong speed.
• Specifically: your chronograph reading. If you had entered a real measured speed, The Lab used it and Guided Build quietly ignored it and used the estimate off your bow's IBO number instead. Drift depends heavily on speed — a slower arrow simply spends longer in the wind — so the two could land a long way apart. On a 65 lb rig with a measured 248 fps, one screen said 3.8 inches at sixty yards and the other said 5.3.
• Spin was worse, on one particular family of vanes. Firenock's Aerovanes do not steer by pushing air off an offset the way a plastic vane does — they generate lift off a rigid three-dimensional shape, which is why Firenock publishes spin figures in the hundreds of revolutions per second. The Lab knew that. Guided Build did not, and quoted single digits for the same arrow. If you fletch Aerovanes and thought that number looked absurd, it was.
• Both screens now call one piece of code for this, not four. That is the actual repair. Every surface that shows you drift or spin — both builders, the arrow diagram, the Shot Solver, the Range confidence model — asks the same function the same question, so they cannot answer differently. There is now a test that fails the build if anyone starts doing the arithmetic locally again, which is how this got in.

FLETCHINGYou can set your helical in Guided Build

• You could always set it in The Lab. In Guided Build you could not, and nothing said so. The control sat on an analysis card that only The Lab draws, so a Guided archer was locked to the two-degree hunting default no matter what was actually on their shafts — while the search box happily offered to take them to "helical" and then dropped them on a screen with no way to change it.
• It now lives with the rest of your fletching choices — straight, 1, 2, 3, 4 or 6 degrees, right next to vane count, in both builders. More helical spins the arrow harder, which settles a big fixed broadhead and kills horizontal flutter; it also costs you a little in wind. The numbers move as you tap, so you can see the trade instead of taking our word for it.

PRE-FLIGHTA shaft change was blaming your vanes

• Change nothing but your shaft, and the check called "Vane drift" would turn amber and tell you your fletching was heavy. Same vanes, same count, same helical. A customer sent us two screenshots of exactly that: he went to a heavier shaft, the arrow slowed about eight feet per second, drift went up three tenths of an inch at sixty yards, and a green check went orange.
• Three tenths of an inch at sixty yards is not something a human being can shoot. The check was grading inches, and inches are mostly a measure of how fast your arrow is going. So it was really an arrow-speed test wearing a vane's name — and the detail underneath it listed his vane, his fletch count and his helical, none of which he had touched.
• It was strict enough to fail its own reference arrow. The whole drift model is measured against a Blazer three-fletch at two degrees. That arrow — the ordinary, correct, nothing-wrong-with-it baseline — got flagged "drift heavy" below 271 fps. Plenty of good hunting setups live under 271 fps.
• It now grades what its name says it grades: your vane, your fletch count and your helical, measured against that baseline. The inches are still on screen, because they are worth knowing — they just do not fail anything any more, and the note underneath tells you when arrow speed is what moved them. Genuinely draggy fletching still gets flagged, at any speed.

FIELD NOTESThe articles match the index now

• Half a job, finished. The last release turned the Field Notes index into a journal and left every article opening in the old layout — centred title, supplier product photo underneath, nothing to do with the writing below it. Open any article now and it is a page from the same notebook: aged paper, a date in handwriting, a drop cap, and the piece itself set in a serif at a width you can actually read.
• The wear is deliberately kept off the words. These are two-thousand-word essays. Aged paper is the kind of idea that photographs beautifully and then costs you something at paragraph forty, so the coffee ring, the fold creases and the dog-eared corner all sit outside the text column, and the writing itself sits on the lightest part of the page in near-black ink with nothing textured behind it. On a phone the ring is hidden entirely, because there it would land on the words.
• The photographs are not deleted, just not drawn. Every article keeps its featured image, so links shared to a group chat still preview with a picture. The article page simply never asks for it. That is a display decision and it can be undone in one line if it turns out to be the wrong one.
• And there is a way onward at the bottom. Previous note, next note, and back to the index — which the old layout never offered, so every article was a dead end.

THE HUNTA front door for a tool nobody could find

• The Hunt has been live since July and unlisted until yesterday. It went into the main menu on the 7th. Before that, nothing on the site linked to it, and the numbers show what that meant: twelve people had ever opened it, and most of those were us testing it.
• The page had no front door either. The tool runs inside a frame, which means a search engine arriving at the page found a title and nothing else — no heading, no text, no links onward. Meanwhile forty pages of real writing about rut timing, moon research and state-by-state breeding dates sat there unreachable from it. The page now has a proper heading and links into all of them.
• What we did not do is rewrite that material onto this page. It would have put two pages on this site competing for the same searches, and search engines resolve that by picking one and ignoring the other. The forty pages were already written. They just needed a door.
• The state list is a menu, not a hard-coded list. Adding a state next season is a two-minute edit rather than a software release, and a state whose page is not ready yet simply does not appear instead of publishing a broken link.

FIELD NOTESA sentence we broke, and how

• We put a sentence in the wrong place in a live article. Adding a line to the saddle-versus-lock-on piece, the insertion landed in the middle of a paragraph — between a claim and the three short fragments that finished it. Those fragments then read as though they belonged to our sentence instead of the author's, which made the passage nonsense. It is fixed, and the tool that does this now refuses to insert anywhere except the end of a paragraph.
• Worth saying why it got through. Every check we had passed: the phrase matched exactly once, the page was valid, and the preview showed the correct sentence being added. What none of them checked was how the paragraph read afterwards. A preview that verifies the edit is not the same as one that verifies the result.

FIELD NOTESThe blog reads like a journal now, because that is what it is

• What you were seeing. Every Field Notes article led with a photograph, and the photographs were supplier product shots — a broadhead on a white background, an insert on a white background. The same broadhead picture ran on two different articles. One insert photo ran on three. They arrived at whatever size the supplier shot them at, from square to a strip barely eighty pixels tall, so the index page was a ragged grid of stock parts photography sitting above writing that is not about parts.
• What it is now. No photographs at all. Each entry is a page from a notebook — aged paper, faint rule lines under the type, the date in handwriting, and a small ink drawing chosen by what the piece is about. An arrow for spine and shafts, a broadhead for broadheads, antlers for the western pieces, wind lines for scent and weather. Nothing on the page is a photograph or an image file; it is drawn in the browser, so nothing can load at the wrong size or fail to load at all.
• Why not just resize the photographs. Because the problem was not that they were mismatched. It was that a stock product photo is a claim about what the article is — and these articles are somebody's notes about what worked, what did not, and what turned out to be wrong. A photograph of a broadhead on a white background says "catalogue." That is the wrong promise to make before the first sentence.
• The articles themselves have not changed yet. Open one and you will still see the old layout. That half is coming.

UNDER THE HOODTwo ways a design can quietly disappear, both closed

• Releases used to be able to eat page designs. Every release rotates between two copies of the storefront, and only files on an explicit list survive that rotation. Anything not on the list reverts — with no error, no broken page, and nothing to notice. The blog design was not on the list. Neither, it turned out, was The Hunt, which has been surviving every release this month by luck rather than design and was one rotation away from becoming a blank page in the main menu.
• The first attempt at the fix was worse than the bug. The list copies files from the live storefront onto the next one, which is right for pages that only exist there. The blog design does not only exist there — it lives in our source code. Copying it from the live storefront would have read the OLD blog design off the live site and pasted it over the new one at the next release, undoing this work while appearing to protect it. Files that come from source are now pushed from source, and files that live on the storefront are carried from the storefront, and the two are no longer confused.
• The release preview now shows this. Before shipping anything we run a preview that lists exactly what the release will do. That preview described the tool build and said nothing about page designs — so it could not have told us whether the blog was in the release or not. It names both now, and it refuses to run at all if a design file has gone missing from source, which is the failure that would otherwise be discovered by a customer.

SHOT SOLVERSix more animals stopped being a deer

• What changed. The Shot Solver draws a photograph of the animal you picked, sized so the kill-zone ellipse sits at its true dimensions on that animal's body. Since v294 it did that for 52 of the 61 species. The other nine got a generic deer illustration with a label admitting it. Six of those nine now draw themselves — the wolf, the coyote, the black bear, the brown/grizzly bear, the polar bear and the Canadian moose. Fifty-eight of sixty-one species are now drawn at a real, cited scale.
• Why they were never drawn, which was not what you would guess. The photographs existed the whole time. Every one of them has been sitting on our CDN since v294, bears and wolves included. A photograph is only usable if we can scale it, and we scale from exactly one number: height at the shoulder, published by somebody we are willing to name. No published height, no scale, no photograph. We would rather show a drawing that says it is a drawing than a photorealistic grizzly at a size we invented. So this was a sourcing job, not an art job.
• The grizzly was blocked because we had been looking in the wrong place. The coastal literature publishes round numbers with nothing behind them — 36 to 60 inches, a 67% spread, no method stated. The interior literature has a real figure: a published mean of 55 adult males five years and older, from Yellowstone, with its own written definition of where the tape goes. We use that one. We deliberately do not blend it with coastal measurements, which are taken to the tip of the claw and run long. A coastal brown bear is a bigger animal than this drawing shows, and we would rather say so than average two things that were never the same measurement.
• The polar bear is the thinnest sourcing in the table and the row says so. Nobody publishes a good shoulder height for them, and the reason is structural rather than sloppy: polar bears are measured lying down under anaesthesia, so the field data is girth, length and skull — never standing height. Every published figure is a sentence in a review rather than a measurement anybody took. We ship the best available and flag it rather than dressing it up.
• The Canadian moose landing on the same 72 inches as the Shiras moose is correct, not a copy-paste. Every eastern agency that publishes a figure converges on six feet at the shoulder. We checked the taller numbers and rejected all three: one is a cow figure, one is boilerplate that also appears on a page about moose imported from Alberta, and one has no citation at all. The two subspecies share a photograph, so matching scales is the honest outcome — previously they drew kill zones four inches apart on identical pixels.
• The wolf and coyote needed a body shape we did not have. Our model turns a shoulder height into a chest depth and a heart-lung height using two ratios per body type, and it had no ratio calibrated on a dog. It does now. The check that it is right: it reproduces the wolf's and the coyote's chest measurements, both set by hand years before this model existed, exactly.

SHOT SOLVERThe coyote was facing the wrong way

• Every animal in this set faces left, and the overlay assumes it. The coyote render faced right. Nothing would have crashed and no image would have looked broken — the kill zone would simply have been drawn on the animal's hindquarter, confidently and at the correct size. It has been mirrored and re-hosted. We found it by drawing the ellipse on every photograph and looking at the result rather than checking that the numbers were plausible.
• Both dogs' kill zones moved forward onto the front leg. The automatic placement puts the vital centre a third of the way back along the body, which is right for a deer and wrong for a wolf — a dog's heart sits tight behind a narrow, upright front leg, not out in the flank. Both were corrected by eye against the photograph.
• The polar bear's kill zone was low. Not against any model — against our own two other bears. Black bear and grizzly put the heart-lung centre at roughly two-thirds of shoulder height; the polar bear was sitting at just under half. It has been brought into line with the other two and checked against the photograph.

SOURCINGThree animals still draw the illustration, and always will

• The wild turkey, permanently. A bird has no withers. "Height at the shoulder" is not a rough figure for a turkey, it is not a figure at all. We did not go looking and we would not use one if it were published.
• The Sitka blacktail, and we went back to check. The agencies that manage that subspecies characterise their deer by weight, dressed weight, hind foot and chest girth — harvest and condition measurements — and never publish a standing height. The only real shoulder height in the literature is a species-wide mule deer figure, and borrowing it would erase the exact size difference the Sitka row exists to represent.
• The Nile crocodile, which needs a different measurement entirely and has no photograph in the set.
• These are refusals, not a backlog. When a species falls back, the panel tells you it is a generic illustration, that it is not this animal, and why. That has not changed and will not.

CLEARED FORThe species list had nowhere to scroll

• What you were seeing. Open Cleared for, try to scroll the list of 61 species, and almost nothing moves. It was reported to us as hard to scroll. On a phone it was worse than that: the list had no height at all. Not one species row was on screen, ever. There was nothing to scroll.
• Why. The panel divides its height between the header, the list, and the sourcing notes at the bottom. Those notes — the paragraphs explaining which floors are Easton's and which are ours — had no ceiling on how much room they could take, and the list had no floor. So the notes took everything. On a laptop they were taking half the panel. On a phone they were taller than the whole panel, which is why the text spilled past the bottom edge, and why the list was left with nothing but its own margins.
• Nothing was cut to fix it. Every word of the sourcing is still there, now behind Where these come from at the bottom of the panel. It opens with one tap and scrolls in its own right, so it can never take the list's room again. The line you see without tapping still says the thing that matters most — that every floor is Sparrow's own number unless it says Easton.
• There is now a search box. Sixty-one species in one column is a long way to travel on a phone even once the scrolling works. Type elk, or buffalo, or africa, or thick-skinned, and the list narrows. The continent and class headings stay pinned to the top as you scroll past them, so you always know which part of the list you are in.
• Two smaller things in the same panel. Dragging the list near the edge of a phone screen used to close the panel out from under you; it now only closes if you actually meant to tap outside it. And on iPhones the Done button could sit underneath the browser's own bar, because the panel was sizing itself against a screen taller than the one you could see.

FINDABILITYIt was only ever in one place, and that place was hidden

• There was exactly one door. Cleared for could only be opened from the final step of Guided Build. If you build in The Lab — the mode most people end up in once they know what they are doing — there was no way to reach it from anywhere on screen. Not a menu, not a tab, not a search result. We found this out because the person who built it went looking for it and could not find it either.
• It now leads the Lethality tab in The Lab. That tab already holds the Penetration Index and the Broadhead Confidence card, and it was built to give the "will it kill" answers one place to live. Cleared for is the most literal of those and it was the one not in there. It sits at the top of that tab now, because which animals your build clears is the question the other two cards are refinements of.
• And it is in The Workshop. The card grid at the bottom of the Builder tab, which exists to show you what this thing can actually do, was listing nine tools and not the one that answers "will this kill an elk". It is there now, in both build modes. Every door opens the same panel with the same numbers — not a second copy that can drift.
• And you can search for it. Type elk, or species, or what can I hunt, or is this legal for cape buffalo, into the search box at the top. It opens the panel where you are, rather than throwing you into the other build mode to get there.

SEARCHSome questions returned nothing at all

• "Will this kill an elk" found nothing. Neither did "check my arrow length." Not the wrong answer — no answer. An empty list.
• Why. The search treats a word as required if any entry can answer it. The word will happened to appear in one buried explanation about collars, and check in one about adhesives — so every question containing either word demanded that the answer also be about collars or adhesives, and nothing qualified. Two ordinary English words were switching off the search. Both are now ignored, the way the and my already were.
• One like it is still there, and we are saying so. Right does the same thing — "is my point weight right" still comes back empty. We have not fixed it because right is also half of how you describe an arrow kicking left or right to the Fix It tool, and switching it off might break the more important question. It is written down, and there is a test that will fail the day somebody fixes it properly.

ACCURACYTwo counts that were quietly wrong

• The group headings were counting from memory. They read North America 29 and Africa 28. The real split is 31 and 30, and 29 plus 28 has never added up to 61. Those two numbers were typed into the labels rather than counted from the list, so they could not notice species being added underneath them.
• We had already fixed this once, one line higher up. The headline at the top of the panel carried the identical error and was corrected last week. The group headings below it were missed. They are all counted from the data now, from one place, so there is nothing left to drift.

Worth saying plainly: this one was reported by somebody who took the trouble to tell us, and it was worse than they described. If something in the Forge behaves in a way that makes no sense, that is worth an email — sal@sparrowexpeditions.com. The measurements behind this fix, on real phones and real laptops, are in the code alongside the change, so the next person to touch that panel can see exactly what broke it.

SHOT SOLVERThe deer that was pretending to be an elephant

• What you saw before. Every animal in the Shot Solver was the same drawing — one deer, for all sixty-one species. A turkey was that deer. An elephant was that deer. The only thing that changed between them was a hidden inches-per-pixel constant, chosen so a fixed 200-pixel kill-zone ellipse would “represent” whatever animal you'd picked. That is backwards: it scales the ruler to the picture instead of the picture to the animal, and it meant the drawing taught you nothing true about the animal actually in front of you.
• What you see now. Fifty-two broadside photographs, each one scaled to its own real body. The vital ellipse is that species' own heart-lung dimensions in inches, and the ranging-error dots sit at their true inch offsets on that animal's body. A springbok and a giraffe are no longer the same picture with a different number under it.
• How the scale is proven rather than asserted. Every photograph is calibrated by hand against a body dimension that is published: height at the withers. Two lines were clicked on each render — the top of the back over the shoulder, and the ground the animal stands on. The pixels between them are a measurement. Divide by the published inches and the whole image acquires a real scale. The photo file stores pixels only; the inches are read at draw time from the sourced geometry table, so there is no second copy of a number to drift out of step with the first — and no uncited figure quietly wearing a publisher's authority.
• Nine animals still show the drawing, and that is the honest answer, not an oversight. Wild turkey, coyote, wolf, Sitka blacktail, black bear, brown/grizzly, polar bear, Canadian moose and Nile crocodile have no published shoulder height we can stand behind. Their photographs exist. What is missing is the sourced number to scale them by, and a photograph scaled against a guess is worse than a drawing that never claimed to be one — it looks precise. So they keep the drawing until someone does the sourcing. Finding the figure is the next job; inventing it is not a job at all.

CHRONOGRAPHThe microphone was never asked for

• The bug. The chronograph did not work on phones. The Forge runs inside a frame on the page, and a frame like that gets no access to the microphone unless the page explicitly hands it over. Ours never did — the permission list mentioned the clipboard and sharing, and nothing else. So the browser refused the microphone before it ever thought to ask you.
• Fixed on both frames, not just the obvious one. The Forge is embedded twice — the full-screen app and the smaller in-page version. Only patching the one you happen to be looking at is how a bug comes back six weeks later wearing a different hat.
• Stated plainly: this may not be the whole fix. The frame is also sandboxed, and browsers can refuse microphone access to a sandboxed frame no matter what permission the page grants. If the chronograph still will not listen, that is the reason, and the answer is a dedicated chronograph page outside the frame — not more attributes piled onto this one. We would rather write that down now than quietly ship a second attempt and call the first one a success.

The photographs sit on permanent storage rather than inside the app, so they are not re-downloaded with every release. Getting them there meant granting the deploy tool a permission it had never held — and the first wiring we tried would have set the picture list on the page outside the frame, where the app cannot see it. That failure would have been silent: no error, no broken image, every animal simply falling back to the drawing while the release notes claimed otherwise. It was caught before it shipped, and the build now refuses to produce an app with no photographs at all rather than let that happen again.

ANIMALSWe were drawing elephants the size of a leopard

• The bug. Every animal in the Forge carries two measurements that drive the hit overlay and the treestand math — how wide the chest is through the middle, and how high the vitals sit off the ground. Not one of the thirty African species had its own numbers. They all inherited a class default, and for cape buffalo and elephant that default was leopard geometry: a 14″ chest and a 22″ vital centre. Giraffe was given a vital centre of 38″, which is below its own brisket. A Nile crocodile — an animal that lies on its belly — was 48″ off the ground.
• What it’s now. Cape buffalo 22″ / 44″. Elephant 59″ / 93″. Giraffe 30″ / 99″. Crocodile 26″ / 8″. Every one of the thirty African species now derives from a published shoulder height with a source you can click, and the model that turns shoulder height into chest and vital position is written down in the open rather than hidden in a constant.
• We checked the model instead of trusting it. Six North American species had numbers set by hand years ago, by eye, before any of this existed. Run the new derivation against published shoulder heights for those same six and it reproduces them within two inches across twelve numbers. That agreement is now a test, so the model can’t drift away from the judgement it was calibrated against.
• And the whole bug class is gone, not just the four rows. The class-fallback tables are deleted. A species with no geometry now stops the build instead of quietly borrowing a leopard’s. You can’t make this mistake again, because there is nothing left to inherit from.
• Where our own data is weakest, it now says so. All thirty-one North American rows are labelled as our numbers, with no publisher behind them. They didn’t change — they’re just honest now, which makes North America the thinner half of the table. That’s the next job. Vital width is still a ratio off vital height for almost every species; we went looking for a published heart-lung width and could not find one for a single animal, so we wrote that down rather than swapping one invented number for another.

LETHALITY“Cleared for elephant” with a 382 grain arrow

• The problem. After yesterday’s rebuild the Forge graded every animal on energy at the bow. For an elephant that is the wrong question entirely — what stops an arrow there is an inch and a half of hide and, on a brain shot, skull. There is no published energy threshold for these animals because energy isn’t what decides it. So they sat at the highest floor we had, which a decent whitetail rig clears, and read cleared.
• Now they’re graded on arrow weight, which is the axis the literature actually uses. The Ashby Bowhunting Foundation publishes weight bands for thick-skinned game: red below 650 grains for cape buffalo, red below 850 for elephant and hippo. That is a real, attributable, quotable number, and it is the only guidance that exists for these animals. A 382 grain arrow is now told plainly that it is 468 grains under a published do-not-use line — and told whose line it is.
• It is ABF’s number, not ours and not Easton’s, and we say so every time. ABF is an advocacy foundation whose stated position is that heavy arrows matter. We cite it anyway, because on these animals it is all there is, and we name it so you can weigh the source yourself. White rhino sits on the elephant band by our analogy — ABF doesn’t name it — so that one warns rather than fails. Nile crocodile we don’t grade at all: nothing published covers it, its vital zone is about four inches, and the barrier is armour and placement rather than bulk. Saying so beats inventing a number and dressing it in a citation.
• The guardrail matters more than the feature. ABF publishes five weight rows and we apply two. Their medium row would put a 400 grain whitetail arrow in the red and their large row a 425 grain elk arrow — mainstream setups that kill cleanly every season, and calling them too light is the worst mistake this tool can make. So there is now a test running 990 comparisons proving the weight gate cannot move a verdict for turkey, deer, pronghorn, hog, bear, elk, caribou, moose, bison or grizzly, at any weight from 250 to 1200 grains. Weight is a lever you pull when you expect to hit heavy bone. It is not a law, and it never becomes one for the animals most of you hunt.

CATALOGFirenock, and the phantoms that wouldn’t die

• The lighted nocks weren’t missing. Firenock indexes its own catalog by style letter, so the nock for a .166 shaft was filed as “G”. If you shoot a .166 you have no way to know that, so it read as absent. Searching 166, .204 or .246 now finds them. Every weight was re-checked against Firenock’s own pages; one was a tenth of a grain out and one crossbow entry was under a style name that doesn’t exist and a grain light. Both corrected, both sourced.
• The phantoms were still leaking. Last month we found catalog entries for products that don’t exist — inserts and collars nobody makes — and hid them. They were still being exported into the store’s own component database through a second door nobody checked. Thirty-one of them, gone. There’s now a test asserting no hidden component can reach that export, so there isn’t a third door.
• Still on the list. Search results for a brand can land in two clusters hundreds of rows apart, which is its own reason things look missing. That’s a sorting fix, and it’s next.

Nobody wrote in about the elephant. We found it while getting our own house in order before asking an African outfitter to look at our numbers — which is exactly the sort of thing you find when you go looking on purpose.

LETHALITYYour elk arrow was probably fine. We were wrong.

• The bug. The Forge was telling people their arrows weren’t enough for elk when they plainly were. A 450 grain arrow at 275 fps — an ordinary, lethal, kills-elk-every-September setup — came back short. In one place it printed the shortfall as “0.00.” If you saw that and thought the tool was guessing, you were right.
• The cause, part one: three engines. Three different screens graded your arrow against two different threshold tables using two different velocities — some at the bow, some derated to 40 yards. The same arrow could pass on the verdict card and fail in the animal matrix. That is not a rounding difference. That is three answers.
• The cause, part two: the arithmetic. The energy minimums everyone in archery quotes come from a chart Easton published, and those are at-the-bow numbers — the figure your pro shop’s chronograph reads. We were knocking a third of the energy off for downrange flight and then comparing against them. That alone made our standard a third stricter than anything ever published. On top of that, Easton’s elk band is 42–65 ft·lb and we had been using 65 — the ceiling — as the floor.
• What we did about the numbers. One engine now, and we publish our own floor table instead of hiding behind somebody else’s: 15 small game, 25 deer and antelope, 42 elk and black bear and boar, 50 moose, 65 the big stuff. Every row says plainly whose number it is. Where Easton names the animal, we say so and quote the row. Where the number is ours, it says Sparrow’s number and tells you what it’s built from. We would rather show you our reasoning and let you argue with it than dress our judgement up as a citation.
• Momentum can no longer fail you. Nobody publishes a per-species momentum threshold. The tables you’ll find online are the energy bands converted with an arrow weight nobody states — circular, and no new information. We still show you your momentum. It will never again be the reason something reads red.
• The mechanical penalty is gone. We were quietly docking mechanical broadheads 15%. There is no source for that figure. The only peer-reviewed study on the question — 1,296 deer, hunters certified and tested annually — found mechanicals recovered better than fixed heads, 90.9% against 82.0%. And by Ashby’s own mechanical-advantage formula, a two-inch two-blade expandable outscores a conventional 1‑1/8” three-blade fixed head. Fixed versus mechanical is not the axis that decides penetration. We stopped pretending it was.
• What we actually believe. A tuned arrow that flies straight is the thing. Ashby’s own list ranks arrow flight second and arrow mass sixth, and kinetic energy doesn’t appear on it at all — he wrote that energy “should be used as neither guideline nor requirement.” His buffalo data includes 196 consecutive hits that broke heavy bone, roughly one in five of them from a 40 lb recurve making about 35 ft·lb. So: energy is a floor, not a grade. Mass is a lever you pull when you expect to hit bone. Get the nock tracking behind the point and then go kill something.
• Where we say we don’t know. A handful of animals have no honest number behind them — nothing published addresses what actually stops an arrow on an elephant or a rhino. Those rows say so instead of inventing a figure.
• Locked down. Twenty real setups — including a 60 lb rig, heavy traditional gear, a youth bow, and Randy Ulmer’s published elk arrow — are now a permanent test. If any of them ever fails its floor again, the build breaks and the message says the floor is wrong, not the rig.

THE FORGEAnd you can just ask it where something is

• There’s a search box now. Type “where do I set arrow length” or “change my broadhead” or “this is too complicated” and it takes you to the control and lights it up — switching tabs, modes and steps on the way if it has to. Two taps from a cold open to your broadhead. It was nine.
• We were speaking the wrong language. The only place to set your arrow length was labelled “Cut Length,” and the broadhead lived under a heading that didn’t contain the word broadhead. If you couldn’t find them, that was us, not you. Both are named properly now.
• A field that ate your typing. The Spine Calculator had a box labelled “Arrow length” that didn’t change your build and quietly overwrote whatever you typed the next time anything moved. It now holds what you type, says out loud that it’s a scratchpad, and offers to snap back to your real build.
• The guided build has a real step bar. Eight unlabelled twelve-pixel dots became named steps you can tap in any order. Point at the one you want.
• The app stops changing itself behind your back. Finishing the guided build used to permanently switch your layout. Tapping one tool in the bottom bar used to permanently turn on every advanced field in the app. Neither happens now — you get what you asked for, where you asked for it, and your settings stay yours.
• Roomier on a phone. The tab strip has its own row again instead of fighting two buttons for space.
• Your saved data was never affected by any of this. Grading, labelling and navigation only — no build, chrono pair or saved setup was changed.

We would rather tell you we got the elk math wrong than keep quietly failing arrows that kill just fine. And people kept telling us the Forge was hard to find your way around — they were right, and “let me show you” is not a fix.

EMAIL + REACHEmail your build to yourself — and the guide pages now have the tool built in

• Email me my build. Drop your email at the capture and you now get a one-time email with a link that reopens your exact arrow build in The Forge — tweak it, print it, or shop the parts that match. One send, not a subscription.
• The guide pages carry the live tool now. The arrow-spine-calculator and how-to-make-a-sight-tape pages no longer just describe the tool — the working builder and sight-tape generator are embedded right on the page, so you can read the method and run it in the same place.
• Easier to find in search and AI answers. Structured data behind the scenes so Google, Bing, and AI search engines read the tools, the articles, and the pricing correctly. No visible change on the page — just a clearer signal to the crawlers.
• Newsletter sign-up, the native way. A clean email-capture band — no third-party pop-up — that files you straight into the store's own customer list.

The tool should meet you where the question is — in your inbox, and on the page that answered it.

RANGE SYNCThe Range tab sync spinner that never stopped — fixed

• The bug. On the Range tab, cloud sync would spin forever and say it couldn't sync — signed in, good connection, still stuck. Nothing was actually saving to the cloud while it spun.
• The cause was the app, not the server. The sync step kept re-triggering itself on every redraw — thousands of times a second — so it never settled long enough to finish. The server was answering fine the whole time; the app just never let the result land.
• The fix. Sync now runs once when you sign in (and on a manual refresh), the way it should. The spinner settles and your range sessions push and pull normally.
• Your data was never at risk. Local saves always worked; this only touched the cloud mirror, and no sessions were lost.

A spinner that won't stop is almost always the app arguing with itself, not the network. This one was.

HOMEPAGEThe front door now opens on the workshop, not the store

• The Forge is the first thing on the homepage now. The old page opened like a gear shop, with the builder as a link you had to go hunt for — the reason most of you are here was three scrolls down. It now leads the page: build an arrow, dial spine and FOC, print a tape, no account and no checkout in the way.
• The rungs are stated plainly instead of implied. Free to build, a free account to save your work so you pick up where you left off, and membership when you're tuning the whole rig — Tune Wizard, sight-tape printing, unlimited builds. The price is on the page, not buried: $8/mo or $72/yr, and the first 100 Founding members lock it in at $36/yr for life.
• The gear you shoot is still right there. Shop-by-category, the hunt-ready bundles, and the "on our bows right now" picks all carried over unchanged — the store didn't move, it just stopped being the first thing a first-time visitor has to wade through to reach the tool.
• Nothing about the builder or its math changed. This is a front-door change only: same spine engine, same dynamic-spine and FOC numbers, same tapes off the same drop curve. If a number was right yesterday it's right today.

Same workshop, same tools, same math — you just don't have to look for the door anymore.

SIGHT TAPEWheel tapes, now exact — validated head-to-head against PCA

• We ran our tapes against PCA's, sight by sight. Three real setups — a linear rail slider, a Spot Hogg wheel, and a Black Gold hunting sight — reproduced from identical ballistics in both tools. Average disagreement: under seven thousandths of an inch, tighter than the tape stock it prints on.
• The Spot Hogg wheel scale got its final digit. A same-bow rail-to-wheel comparison across 246 yardage gaps pinned the wheel's true stretch at 1.677 (± 0.003) — our previous 1.66 came from measuring printed PDFs and ran about 1% short at long range. Boss Hogg, Tommy Hogg, and Hogg Father tapes all carry the corrected scale.

SIGHT TAPEPick your sight — the tape sizes and scales itself

• No more comically wide tapes. Every sight in the catalog now carries its own tape width, and it applies the moment your setup's sight is selected — no preset button to find. Modern sights run narrow channels, so the fallback width dropped from 0.50" to 0.25" to match. Measured your slot with calipers? Your saved width always wins, and the setup card tells you when a width was auto-sized versus hand-set.
• Wheel sights print at wheel scale. The Spot Hogg wheel family — Boss Hogg, Tommy Hogg, Hogg Father — shares one elevation mechanism with the Hogg Father Pro, so all of them now print at the Pro's verified 1.66× scale instead of a too-short linear strip. The Tape scale row in the setup card shows the factor and where it came from.
• Inverted sights print backward — correctly. Some spool and pendulum sights run their indicator UP the tape as range grows (HHA Optimizer Lite series, Apex Covert, Viper Sidewinder, and friends). Their tapes now come out mirrored to match, across the preview, PDF export, label printer, and gang print alike.
• Every path agrees. Preview, PDF, roll-printer, label image, travel tapes, and gang print all resolve the same per-sight width and scale — what you see on screen is what lands on the sight bar.

DESKTOPThe live stats pill comes to the big screen

• Your numbers follow you on desktop now too. The live stats readout that shipped on mobile in v278 earned a spot on the big screen: scroll deep into a build and a glowing pill anchors bottom-left with your spine verdict, speed, finished weight, and FOC — updating with every component you touch.
• Click it for the full verdict. A popover opens right above the pill with the complete picture — energy, penetration, species checks, all of it — without yanking you back to the top of the page. Click anywhere else or hit Escape and it's gone.
• It knows when to disappear. Scroll back up to where the full stats panel is visible and the pill bows out. One brain drives both the mobile bar and the desktop pill now, so the numbers can never disagree.

MOBILEThe Lab on your phone: live stats that follow you down the page

• No more scroll ping-pong. Building an arrow on a phone meant picking a component, scrolling back up to see what it did to your numbers, scrolling back down, repeating. Now the moment your verdict panel leaves the screen, a slim bar slides up from the bottom with the numbers that matter — spine verdict, speed, finished weight, and FOC — updating live with every change you make.
• Tap the bar for the full picture. The whole verdict slides up as a sheet — speed, kinetic energy, spine, penetration, weight, FOC, and the species checks — then gets out of your way. The bar also knows when you can already see your stats and stays hidden.
• Desktop untouched. This is phone-and-small-tablet only; the full spotlight panel works exactly as before everywhere else.

EXPERIENCESimpler menus — the gear is now one tap away, everywhere

• The Forge's menu got out of its own way. The in-app menu used to mirror the entire site navigation — every tool listed twice, and the store buried at the bottom. It now leads with Jump to Tool, then a tight Shop gear section: the store, bundles, arrows, and brands. Two taps from any tool to any product.
• The Shop menu is now a wall of gear, not a list of words. Hover Shop anywhere on the site and you get the full category grid — bundles (with real savings called out), arrows, rests, sights, releases, broadheads, and the rest — each a tile, not a text link.
• Less clutter in the header. Workshop pages like My Bows, My Arrows, and the Range Log live inside The Forge where they belong, instead of crowding the site menu.

ACCURACYSight tapes: the long-range bend is now real physics, not a fudge factor

• A user printed the same build on our tape generator and on Precision Cut Archery's, and the marks disagreed by a third of a yard of dial at 100. We pulled the actual printed mark positions out of both PDFs and traced it: past your far chrono reading, our drop model leaned on a single calibration constant that fit one decay rate and drifted on steeper ones. Wrong answer for the right reason — so we replaced the constant with the mechanism it was imitating.
• The drop integrator now brakes the arrow's fall with drag, the way the real world does. Drag doesn't just bleed forward speed — it also resists the growing fall speed, which is why real trajectories bend flatter at distance than a bare gravity calculation says. That effect now comes straight out of your own chrono numbers instead of a one-size-fits-all factor.
• Validated against six PCA exact-mark references — including the tape that caught us. Every mark from 20 to 177 yards now lands within 0.02" of PCA's field-proven prints, most within 0.005" — a fraction of a 0.019" pin. The build that showed the drift went from 0.010" off at 140 yards to 0.003". These references are locked into the test suite; the math can't quietly regress.
• Every tool now shares one drop curve. The tape, the drop table, ranging forgiveness and the Shot Solver used to split hairs downrange; they now integrate the identical trajectory, so the number on your printed tape is the number every other tool reasons from.

Chrono readings stay sacred: your measured speeds are reproduced exactly at the distances you measured them. This change only sharpens what happens between and beyond them.

CHANGEDOne membership. Founding Flock: $36/yr for life, first 100 members

• Tiers are gone. Hunter and Pro made you guess which tools you'd need before you'd ever used them. That was our problem, not yours. Now there's Free, and there's the membership — every tool on the bench, everything we ship next, one price.
• $8/mo or $72/yr — and the first 100 members lock $36/yr for as long as they stay subscribed. Yes, that's exactly what the single-purpose calculator apps charge. On purpose. Run The Forge next to whatever you use now for a season and decide which one made you better.
• Nobody loses anything. Existing subscribers keep everything they had and gain everything they didn't. If your plan said Hunter, you're a full member now.
• Coming: Beginner / Standard / Expert views. They'll change what's front-and-center so the bench isn't overwhelming on day one — never what you're allowed to touch. No tool will ever hide behind an experience level.

Software costs us almost nothing to serve, so the price is a statement, not a margin calculation. The statement: we'd rather have you in the flock now than protect a margin we don't need yet.

NEWAsk Sparrow — ask the shop anything, get an answer backed by data

• There's a new button in the corner of every tab: Ask Sparrow. Quietest vane? Best broadhead for elk? What to do about a tail-left tear? Ask in plain English. Answers come from the same verified data the builder runs on — the sound-tested broadhead numbers, the sourced KE/momentum minimums, the vane study results — plus your live build if you have one loaded. When a number isn't in our data, it says so instead of making one up. Every generic chatbot can bluff an archery answer; this one doesn't have to.
• It hands you the tool, not just the words. Ask about a paper tear and you get the fix and a chip that jumps you straight into the right Tune step. Answers are guidance — the builder's math is still the source of truth, and Ask Sparrow will tell you that itself.
• Free to use: 3 questions a day anonymous, 10 with an email, 100 on Pro.
• We also corrected four tuning directions in the Tune lab. The bareshaft and broadhead pages had their rest-adjustment directions flipped relative to the paper-tear pages, and the walk-back drift tip pointed the wrong way (drift at distance = move the rest opposite the drift, then re-zero). All surfaces now teach one consistent system, left-handed shooters get an explicit mirror note, and paper tuning now says what we actually teach: start at 6 feet, finish with a clean bullet hole at 12 — that's the distance that proves the tune.

Yes, we shipped an AI thing. It only knows what we can back with sources or our own testing, it cites them, and it refuses to freelance the spine math. That's the only version of this we were willing to build.

FIXED1Arrow vane picks now stick

• Fixed a bug where choosing a 1Arrow single-piece vane snapped back to your previous vane. Now it selects cleanly, locks the count to 1, and switching to a normal vane restores a 3-fletch count.

NEW1Arrow single-piece fletch

• 1Arrow's 1Vane and 1Wing are in the vane list. The 3D-printed wrap-and-vane that installs as a single piece but flies like a four-fletch.
• Real per-piece weights, straight from 1Arrow. 1Vane 23–24 gr, 1Wing 22–23 gr — corrected from what we had before.
• Fletch count locks to 1. Pick one and your arrow weight and FOC come out right, no extra math.

One piece on the shaft, four-fletch behavior downrange — and the build math finally reflects it.

NEWDial your FOC up front

• Screw-in front weights, right in the point section. Add-on weights — Gold Tip FACT, brass, and steel — now live in the point picker. Stack grains at the tip and watch your FOC climb live as you build.
• No more phantom parts. Pulled a handful of components no manufacturer actually sells, so every option in the picker is a real part you can buy.

Front-of-center is where a broadhead earns its keep — now you can chase it without leaving the build.

IMPROVEDChecked against the factory

• Every spec re-verified. We took each major shaft, insert, nock, vane, and broadhead and checked it against the manufacturer's own published numbers.
• Corrections across the board. Fixed spine and inside-diameter figures, insert and nock weights and materials, and broadhead cut diameters wherever they'd drifted.

If the picker says it, the factory says it too.

FIXEDStraight from Victory

• Victory VAP TKO insert, corrected. It now reads SHOK TL at 50gr in 7075 aluminum, matching Victory's published spec.

One shaft, one number, made right.

IMPROVEDThe Armory moved to the front of the shop

• Armory is now the second tab — right beside Builder, visible on a phone without scrolling the tab strip. It's also in the storefront menu under Jump to Tool. The rack shouldn't be in the back room.
• The bottom bar lit up. The phone quick nav (Build / Range / Solver / Armory) traded camouflage for ember — orange icons, and the tab you're on becomes a solid ember pill. If you didn't know it was there, you do now. 1,344 passing.

Round three of "make it one tap" in a single evening — each one from real thumbs on a real phone.

FIXEDThe quick nav no longer hides from beginners

• Guided Build and the quick nav now share the bottom of your phone. v252's quick bar politely stepped aside during Guided Build — which meant the people who live in Guided Build never saw it at all. Backwards. Now the Next-step bar stacks directly above Build / Range / Solver / Armory, so the next step AND the whole workshop stay under your thumb at the same time. 1,344 passing.

Caught by Sal on his own phone within the hour. That's the loop working.

IMPROVEDEverything under your thumb

• Mobile quick nav. A slim bar now rides the bottom of the screen on phones: Build, Range, Solver, Armory — one thumb-tap each, no scrolling for tabs. Standing at the line? Range is right there. Wind picks up mid-hunt? Solver. It politely steps aside whenever Guided Build or a live session needs the bottom of your screen.
• The Armory ends with the truth. After every number on the rack has been weighed, the last thing you see is our sticker: shoot your f···king bow. Tap it — it opens the Range. The punchline is a working button. 1,344 passing.

Data in service of reps, never instead of them.

NEWThe Armory — which arrow rides today?

• Every build, on one rack. The Saved Builds tab is now the Armory: each of your arrows as a card — weight, speed, KE, momentum, FOC, spine — with whitetail / elk / moose clearance badges, your range history for that setup, and an honest chip that says whether those numbers are MEASURED from your chrono and scale or still engine estimates. It's the night-before-the-hunt question, answered at a glance.
• One tap to work. Every card carries Build / Sight tape / Range buttons — pick the arrow and you're already in the tool with it loaded. Hunter and Pro members get the full rack; everyone can walk through the live preview. 1,344 passing.

Born from a custom build sheet we made for a friend. It was too good to keep private.

IMPROVEDShare links, hardened and gift-wrapped

• A shared build link is now a promise. The link format is frozen under a permanent compatibility test — any future change that would break a link you've already sent someone fails our build before it ships. Send a build today, open it in five years.
• Links can carry a gift. Some links you receive from us or our partners now unlock the full Pro experience the moment they open — no account, no card, no gates. If someone hands you one, that's not a bug. Enjoy it. 1,344 passing.

Something's coming. The carpet is down.

NEWYour dozen gets names, your data gets a memory, and shared builds get a front door

• The Dozen Tracker — find the arrow throwing your flyers. Number your arrows, flip on "Track individual arrows" in a live session, and shoot in order — the number advances itself, zero extra taps. After enough tagged shots, the new Trends page issues a per-shaft verdict: "Arrow #7 is 40% looser than your average — spin-test it." It also names your hammer, the shaft that centers every group. That one goes first on opening morning.
• Trends — your whole season on two charts. Every chrono import you've saved becomes a speed-over-time line with an SD band (watch a new string settle; catch serving wear before it costs you). Every logged session becomes a group-size trend at each distance, with a straight answer on top: tightening or opening, and by how much.
• Shared builds get a real welcome. Open a build link a buddy sent and the Forge now greets you with the arrow itself — name, weight, speed, energy, FOC — and one button: Open this build. No questionnaire standing between you and the arrow someone wanted you to see.
• Under the hood: 408 bow pages, computed honestly. The engine can now generate a recommended starting build — spine, weight, speed, energy — for every bow in the catalog, each with a link that opens the exact build in your Forge. Rolling out to search over the coming weeks. 1,332 passing.

An arrow is guilty until proven innocent — now there's a trial.

IMPROVEDThe tour took its blindfold off

• The guided tour no longer blurs out the workshop. It used to dim and lock the whole screen while it talked at you — you were reading about a page you couldn't see. Now the tour is a small card in the corner (bottom of the screen on your phone) and the page behind it is fully live: scroll it, tap around, try the tool it's describing, then hit Next when you're ready. See one tool per stop, hands on, all nine stops. 1,326 passing.

A tour should show you the shop, not stand in front of it.

IMPROVEDThe next step finds you now

• Guided Build's next step is always on screen. The Back / Next buttons used to live at the bottom of the step — below the fold on most screens, easy to lose. They're now pinned to the bottom of the window on desktop and phone alike, with the step counter and the name of what's next ("Next: Pick shaft") right on the bar. Finish a choice, hit the button that's already under your thumb. 1,326 passing.

If you have to hunt for the next step, the wizard failed. Now it can't hide.

NEWFix It, Beginner mode, and a workshop that finally explains itself

• ✚ Fix It — tell us what your arrows are doing. New button, top of every page. Pick the symptom in plain words — hitting left, groups falling apart past 40, broadheads flying different, random flyers, loud bow, lost speed, weak penetration, punching the trigger — and you get the diagnosis and the ordered fix, free, right there. When a guided repair exists, one tap drops you into it. If your bow has a problem, the Forge now has the door.
• Every term explains itself now. Tap any dotted-underline word — spine, FOC, kinetic energy, IBO, let-off, GPI, brace height, twenty-five of them — and get the plain-English answer in two sentences. Then tap "the nerd version" for the formula and the sources. Beginners get clarity; archnerds get receipts. Same tap.
• A real Beginner mode. The old "Simple" toggle quietly showed you almost everything anyway — we called ourselves out on it. There are three honest levels now: Beginner (build an arrow, see if it works, log your shooting — no jargon walls), Standard (the everyday workflow), and Expert (every field we've got). And if you don't know your draw length, the first screen now measures it from your arm span.
• Diagnosis is free for everyone. The Range tab's automatic group diagnosis and the live spine gauge are no longer behind the paywall. Knowing what's wrong is free; the step-by-step guided cure is what Hunter buys.
• Your chrono data is remembered, all of it. Imports now keep every shot — not just the average: standard deviation with an honest verdict (under 2 fps: excellent; over 4: here's what to check), extreme spread, the full LabRadar decay curve, and a speed history that shows you when your strings are settling. Plus one-tap export of all your saved builds.
• Share from the moment that matters. The verdict card and Proving Grounds results now copy a link that opens your exact build in anyone's Forge. 1,326 passing.

Today's standard was simple: a first-timer should never feel stupid here, a hardcore tinkerer should never hit a ceiling, and nobody with a problem should have to guess which tool holds the answer. Grilled our own work against that bar and rebuilt what failed. Built, not bought.

NEWThe Proving Grounds — test your own gear with your own arrows, and drop your chrono file while you're at it

• One build. One variable. A real verdict. New in Range: lock your entire arrow except the piece you're questioning — vanes, broadheads, point weight, or nocks — shoot the same short protocol with each candidate, and get a measured answer instead of a forum argument. The base build is frozen the moment the test starts, so nothing drifts underneath your data.
• Broadhead and lighted-nock tests answer to a control. Tag your field point or your stock nock as the baseline, and the verdict speaks in the terms that matter: "1.1-inch left of the control at 60 — expect a rest adjustment if you switch." Group centers, not just group sizes.
• The verdict doesn't lie to you. When one candidate holds speed and another groups tighter, we don't fake a winner — the table shows both, and a hunting ⇆ target toggle explains its preference either way: retained energy and drift for the hunt, pure sigma for the X-ring. Group sizes are count-normalized (a 3-arrow group and a 10-arrow group compare honestly), and a run only counts as complete with a chrono reading plus groups at two distances. Partial runs are labeled, never averaged away.
• Drop your chrono file. Garmin Xero owners: export CSV from ShotView, or plug the unit in and grab the FIT file — we read Garmin's own format shot-by-shot, using their official SDK. LabRadar owners: your CSVs carry downrange speeds, so one file fills both chronograph readings — launch and distance — and switches on dual-velocity sight tapes in a single tap. Everything previews before it applies; nothing is set silently.
• Where this goes: completed test runs will soon feed an opt-in, anonymous dataset — Forge Field Data — that tightens the drag and drift models for everyone, on top of the PCA 2025 study baseline. Your range day, everyone's better numbers. 1,326 passing.

This one's been the dream since day one: the tool that doesn't just predict, but hands you the protocol to check the prediction — and gets smarter every time somebody does. Test your vanes. Test your heads. Test that lighted nock you're suspicious of. Built, not bought.

NEWYour scale is the referee now — plus room for the 90-pound bows

• Draw weight now goes to 100 lb. The sliders topped out at 90 right as the AX-90 class showed up — and field samples of those bows come in hot, some near a true 100 at the wall. Every draw-weight input now runs to 100. The spine engine already handled it: the stiff-end chart extension from v235 covers effective weights past 100 with the validated 150–200 floor, so the verdicts up there are the same chart-grounded calls, not extrapolation.
• Glue is real weight, and now it's counted. A user weighed his finished arrows and caught us 2–4 grains light — insert epoxy and fletch adhesive that a dry component sheet ignores. New "Adhesives & scale check" section in the builder: epoxy and fletch glue as honest, position-aware line items (epoxy up front nudges FOC up, fletch glue in back nudges it down — physics, not padding). New builds start at a typical 2.5 gn; your existing saved builds don't shift a single grain unless you set it yourself.
• Weigh the finished arrow and enter it. The new scale field works exactly like a chrono reading does for speed: your measured grain weight overrides the sheet for kinetic energy, momentum, and the speed estimate, with the sheet-vs-scale delta shown right there. If the delta runs past ±6 grains, we say so — that usually means a component's listed weight is off, and now you'll catch it.
• Spine math untouched, again. Adhesives are deliberately excluded from the spine charts — the published charts were built from glued arrows, so counting glue twice would be wrong. Easton/Victory parity (98% / 89%) pinned by a dedicated test. 1,301 passing.

Predict, then measure, then trust the measurement — that's how the chrono field has always worked here, and now the grain scale gets the same respect. Thanks to Cody for weighing his arrows and telling us we were light. That's the loop working. Built, not bought.

NEWThe 60-second hook — your bow first, the verdict second, and every tool finally visible

• First visit starts with one question: what are you shooting? New visitors used to get a legal wall, then a subscription pitch stacked on top of it, then an empty workshop. Now the first screen is a bow picker — brand, model, draw weight, draw length, pulled from the catalog — and the next screen is your arrow graded live on your bow. The safety acknowledgment didn't go anywhere: it's the same acceptance, recorded the same way, folded into the button you were going to press anyway, full text one tap away.
• The Verdict Moment. Speed, kinetic energy, spine verdict, and penetration for your actual setup — plus the animal-qualification badges (Whitetail / Elk / Moose, DNR and Pope & Young sourced). And when a light-but-fast arrow clears the energy bar but falls short on momentum, the badge says what actually fixes it — add arrow weight — instead of quoting an irrelevant ft·lb number.
• The Armory. All twelve tools, laid out as cards at the bottom of the builder — what each one does for you, which tier it lives in, one tap to open. Locked tools open in full live preview; nothing is hidden behind a toggle anymore. The tool grid renders from the same registry as the tab bar, so it can never drift from what actually exists.
• Upgrade CTAs stopped eating your session. Every "unlock" button used to navigate the whole page away to the pricing page — killing the build you were in the middle of. Pricing now opens in a sheet inside the tool; only actual checkout leaves for the store.
• Share links are live. Every build, range session, and shootout can now be shared as a link — the recipient lands on a branded page and one tap drops the exact build into their own Forge, free, no account. The share cards carry no name, location, or photos unless you turn them on.
• Spine math untouched. Easton/Victory chart parity (98% / 89%) and the v235 dynamic-spine physics stand exactly as they were. 1,290 passing.

v242 fixed what the tool computed; v243 fixes what a first-time visitor sees. Same workshop, same math, same free tier — the difference is that the thing The Forge does best now happens in the first minute instead of hiding behind three walls. Built, not bought.

BUG · fixedLive speed — the fps readout finally tracks your build, and a New Setup is actually new

• The speed stat was frozen at 285 fps — for everyone, on every edit. The Live-build bar's Speed number was seeded at 285 and never recomputed: change your draw weight, your IBO, your point weight, your whole bow — it sat there. Worse, kinetic energy and momentum were quietly computed from that same frozen number, so a 480-grain build out of an 80-lb bow and a 380-grain build out of a 45-lb bow reported KE off the same 285. Now the speed re-derives on every edit — measured chrono reading if you've entered one, honest IBO-based estimate if you haven't — and KE/momentum move with it.
• Number fields you can actually clear. The two-reading chronograph fields (and a dozen other numeric inputs across the tool) would snap back to a stuck 0 the instant you tried to blank them — you'd fight "040" trying to type 40. Every numeric field now holds what you're typing and commits when you leave the field, blank means blank, and the chrono block has a one-tap Reset that clears both readings.
• New Setup no longer smuggles in your last build. Loading or resetting a build was a partial overwrite — anything the incoming build didn't specify (a chrono reading, a custom insert weight) silently survived from the previous one. Load and reset are full replacements now: the next customer's build starts from zero, and only pulls in what you explicitly load.
• Deleted rigs stay deleted. Delete a rig, reload, and it walked right back in — cross-device sync treated "missing from the list" as "not synced yet" and helpfully restored it. Deletes now leave a tombstone that syncs everywhere, so a deleted rig can never be resurrected by a merge, a reload, or another device.
• Spine math untouched. No changes to the spine engine — the Easton/Victory chart parity (98% / 89%) and all v235 dynamic-spine physics stand exactly as they were. 1,286 passing.

All four of these came from one build session — setting up a customer's arrows start to finish and hitting every rough edge in a row. The frozen 285 was the humbling one: the recompute existed, tested and correct, in a code path the app stopped using months ago. The fix wires it into the path that actually runs, with regression tests so it can't silently detach again. Built, not bought.

BUG · fixedMeasured launch speed — type your number without it jumping on you

• The bug: the optional “measured launch speed” box in Bow Setup snapped to its 120 fps minimum on every keystroke — so a number on its way in (“3” heading toward “300”) jumped to 120 before you finished, and you couldn't type the speed you actually chronoed.
• The fix: the field now keeps exactly what you type and only snaps into the valid 120–420 fps range when you click away or press Enter.

A small one, but the kind that quietly makes a tool feel broken. Flagged it, fixed it same hour.

Rated IBO vs Real-world speed — your spine verdict on the speed your bow actually makes

• A speed-basis toggle in Bow Setup. IBO is the factory's lab number — every bow clocked the same way (70 lb, 30″ draw, 350-grain arrow, bare string) so you can compare brands fairly. It's the EPA mpg on the window sticker: a real, standardized number nobody actually drives. Bolt on a peep, a D-loop, and vanes and shoot it at your draw, and you're off the bench — the number settles a little lower. The Forge now shows both, and defaults the spine verdict to the real-world number.
• Both numbers, always. The card reads e.g. 353 fps rated → ~340 fps as you'll actually shoot it, using a per-bow chrono measurement where we have one (38 flagship bows tested) and a measured blanket gap where we don't. Nobody fudged anything — you're just not shooting the test.
• Why it matters: a genuinely fast bow could trip the speed-step penalty on the spine chart and read your arrow “slightly weak” when it tunes perfectly in the real world (thanks Cody). Real-world mode puts those builds back in range. Speed, KE, momentum, and trajectory already ran on your real numbers and are unchanged; one shared resolver feeds every spine surface, so the gauge, checklist, what-if, recommender, and Build Review can never disagree.
• Arrow wrap, for everyone. The wrap option used to hide in the advanced view. It's a basic choice now — available at every complexity, off by default, and weighted into FOC at the tail when you turn it on.

The spec sheet is honest and useful — it just isn't the shot you take. We show you both: the number to compare bows, and the number to tune yours.

BUG · fixedTAC vanes are back in the picker — plus wind drift for every TAC size and a build-sheet order fix

• TAC vanes weren't showing up. The full TAC lineup — Driver, Matrix, Summit, and the High Profile broadhead vane — was in the catalog, but the live workshop was running a build from before that audit landed, so none of them appeared as choices. All ten sizes are selectable again.
• Wind drift, modeled per TAC size. Only two TAC Drivers had measured drift numbers; the other eight fell back to a generic default. Every TAC vane now carries its own wind-drift coefficient, anchored to the two PCA-measured Drivers and scaled by length and profile height. Estimates are labeled as estimates, measured values as measured.
• Sketch builder back in the right order. The component list in Sketch mode had slipped back to an old arrangement. It now follows the arrow from back to tip: bow, sight, arrow, nock, wrap, vanes, insert, collar, CTI, point.

The vanes were never lost — the live build was simply a step behind the catalog. Shipping this release is what put them back on the shelf.

BUG · fixedMembers stopped getting re-onboarded — no more “pick a tier” or “join the list” after you've already paid

• The bug: signed in, already subscribed, and The Forge still popped “Pick your tier” and an email-capture box on load — as if you'd never been here. Exactly the kind of thing that makes a paying member wonder whether their payment even took.
• Root cause: the builder loads inside Shopify and spends the first moment not yet knowing who you are. Both first-visit prompts fired in that blind window, keyed off a “have you seen this” flag that never checked whether you were already a member.
• The fix: both prompts now wait until your account and tier are confirmed, and never show for a signed-in member or any paid tier. If your account confirms a beat after the page opens, the tier prompt closes itself. Regression-locked. 555 passing.

A member should never have to wonder whether the thing they paid for knows they paid.

BUG · fixedOne physics — a deep audit of every engine, and everything it caught

• One velocity model — for real this time. v166 said "one velocity model everywhere." A deeper audit proved it wasn't: the trajectory table, the sight-tape fallback, the shot solver, and the express tape each still carried their own decay law, and the trajectory card showed roughly 20% more drop than the validated sight tape for the same arrow. There is now literally one velocity module every panel derives from, regression-locked so they can never disagree again. With a chrono reading, it uses your measured decay; without one, it uses an estimate anchored to our measured chrono data — and it's labeled as the estimate it is.
• One wind model too. The Shot Solver's windage came from a different wind engine than the Build Review drift card — they could disagree by up to 70% on the same shot. The solver now reads the same calibrated wind-drift engine as everything else.
• Lethality graded at 40 yards. Game-class verdicts (Kill-Zone card, Animal Matrix) were graded on muzzle energy while the Penetration Index graded at 40 yards downrange. Everything now grades at 40 yards, and the elk threshold is aligned to the published 55 ft·lb guideline. Some marginal builds read tougher now. They should.
• Crash fixes. Clearing a step field in the Cut Chart could freeze the tab; the Range Log's "Open Tune step" link could error out the Tune tab entirely (and quietly skipped the Hunter-tier gate — closed); the wizard's "Run the Shot Solver" and "Print sight tape" buttons silently bounced back to the Builder.
• Every input survives being blanked. The June 12 wind-drift white-screen turned out to be a pattern, not a one-off — clearing any numeric field could feed a zero into the math. Every input in every tool now keeps its last good value instead, and saved setups are sanitized when they load.
• Things that silently didn't save, now save. Tune wizard progress and before/after history, your email (so we stop re-asking), and first-visit flows now persist properly on the live site through the same storefront mirror as your builds.
• Share buttons hidden for now. They produced links to a share page that doesn't exist yet. They return when the page does.
• Tabs reordered to match the workflow: Builder → Tune → Sight Tape → Range → Shot Solver → Saved Builds → Spine → Recommender → Physics → Database. 552 passing.

Two subscribers found bugs this week. That stings in the right way — this release is the full-codebase audit it triggered: three independent sweeps (inputs, math, wiring), thirty-plus findings, every one fixed or publicly listed. The audit doc ships in the repo.

BUG · fixedHonest save status — the “storage full” warning that wasn't

• Fixed a false alarm: editing a setup in The Forge could pop “Device storage is full — changes can't be saved.” Storage wasn't full. Shopify serves the builder inside a security sandbox where the browser rejects every save to the frame's own storage, and we misread that rejection as a full disk. Your changes were being saved the whole time — by the storefront-page mirror that has carried your builds and setups since v164. The warning now only appears when storage is genuinely full.
• Range Log hardened: sessions logged inside the embedded builder previously lived only in the sandboxed frame's storage — which the browser doesn't keep. The Range Log now persists through the same storefront-page mirror as your builds and setups, and restores from it on load.
• Regression-locked: six new tests pin the difference between “storage is full” and “the sandbox said no.” 514 passing.

Caught while filming launch clips: the scary toast fired on the first slider drag. A warning that cries wolf trains people to ignore the one that matters.

One spine brain

• The build recommender now derives its spine filter from the same canonical engine as the spine gauge, pre-flight checklist, and what-if matrix — its private duplicate math (which still used the pre-recalibration coefficients and ignored insert weight) is deleted. A 768-case automated sweep guarantees the recommender can never suggest a spine the spine check then flags.
• Heavy-end recommendations shift appropriately weaker in line with the June 11 chart recalibration; light and mid-range picks are unchanged.

Chart-true spine — the engine recalibrated to three manufacturers' published charts

• Recalibrated to manufacturer chart data. After validating the spine engine against Victory's, Easton's, and Gold Tip's published charts (437 chart cells): the length correction is now 3.5 lb/in (was 5), front-mass is 0.12 lb/gr with the first 25 grains of insert weight exempt (charts already assume a standard insert), and IBO speed is rebanded so 301–340 fps is the baseline. The engine now reproduces Easton's chart in 98% of cells and Victory's RIP TKO chart in 87%.
• Real-world anchor. A 70# / 29" / 175gr-front build on a 300 spine — Victory's own chart pick, and a build our shop tunes routinely — previously read TOO WEAK; it now reads in-window. Full methodology and divergence tables in the repo (SPINE-CHART-VALIDATION-2026-06-11).
• Regression-checked from here on. The validation harness ships with the repo, so future engine changes are regression-checked against all three manufacturer grids.

BUG · fixedTool-page sync fix — the standalone tools finally know who you are

• Fixed: My Bows, My Arrows, Tune, and the other standalone tool pages never received a signed-in identity from the page wrapper — they always ran as anonymous, so gear saved there stayed on one device and never synced phone↔desktop. The wrapper now answers the identity handshake (reading it lazily at message time — the v167.1 lesson).
• Your saved gear comes along. If a bow or arrow profile you saved before this fix is still on the device where you saved it, opening that tool signed-in will now mirror and cloud-sync it automatically.

BUG · fixedSpine truth — one verdict everywhere, plus Load from My Bows / My Arrows

• One spine engine, three surfaces. The Wizard pre-flight "Spine match" check now uses the exact same spine engine as the spine gauge and the what-if matrix. Previously a simplified heuristic could call a build "stiff · tune-tolerant" while the gauge read TOO WEAK for the same arrow (thanks Austin for the catch). All three surfaces now agree, with a regression test sweeping 576 input combinations.
• New: Load from My Bows / My Arrows. Saved gear profiles from the My Bows and My Arrows tools can now be pulled straight into the Wizard and Sketch builders — one tap, no re-typing. Works locally and, when signed in, across devices via cloud sync.
• PDF spec sheets, sharpened. They now lead with THE FORGE — powered by Sparrow Expeditions, include a spine readout (selected spine, chart window, and verdict), and say whether the velocity behind the KE number was measured or estimated.

BUG · fixedCloud sync actually syncs — three stacked bugs, found by testing the live site

• The honest version: v166 said "your setups follow you across devices." The code shipped, but a live end-to-end test the same evening showed nothing ever reached the sync server — for any signed-in user, ever. Not one bug; three, stacked, each silently masking the next.
• Bug one: the sync server's security allow-list was missing one of the three identity headers every request carries, so browsers refused to send anything at all.
• Bug two: Shopify serves the builder inside a security sandbox that strips its web origin; the sync server rejected those "originless" requests. Now explicitly allowed — identity is verified cryptographically, not by origin.
• Bug three, the deep one: the storefront page answers the builder's "who is signed in?" question — but it captured that answer at page load, one script too early, before the sign-in data existed. It has answered "anonymous" to every visitor since the bridge shipped. Subscription tiers read the data later, which is why paid unlocks worked while sync didn't.
• Verified live this time: a setup saved in the builder now lands in cloud storage and follows your sign-in to any device. Range Log cross-device sync — believed working since v136 — engages for real now too. 467 passing.

A test suite can't catch a bug that lives between four systems — the storefront page, a sandboxed iframe, a CORS preflight, and a worker. Testing the real site, signed in, on the live theme caught all three in one evening. That's the standard going forward.

BUG · fixedDeadly accurate — one physics engine, audited end to end

• One velocity model everywhere. An internal audit caught the lethality tools (Penetration Index, game-class confidence) using a simpler speed-decay model than the trajectory and sight-tape engines — optimistic by roughly 9 fps at 40 yards and 18 fps at 60. Every tool now reads the same calibrated decay law, with a regression test so they can never drift apart again. Some marginal long-range verdicts will read slightly tougher now. They should.
• Spine recommendations validated against the manufacturers' own charts. The spine engine now runs against 22 cells of Easton's Hunting Shaft Selector and 10 cells of Gold Tip's compound chart in the test suite — mean error about 4 spine units, every cell within half a spine class. The validity range (compound, 250–500 spine, 26–31" arrows, 75–150 gr points) is documented instead of implied.
• Per-spine diameters for 33 shafts. A 300-spine and a 500-spine of the same model aren't the same diameter — wind drift, penetration, recommendations, and competition compliance now use your actual spine's OD instead of one nominal number per model.
• Range Log → Tune now lands on the right step. Diagnose cards said "Open Tune step" but dropped you wherever the wizard was last open. Fixed.
• Competition shaft-OD check is real now. It previously always passed as an advisory placeholder — a false green that could cost you at equipment check-in. It now compares your shaft's OD against the division limit and warns when you're within .005" of it.
• Your setups follow you across devices. Sign in on your phone and your bow setups are there too, not just your Range Log.
• Honest-copy fixes: the Hunter tier card now states its real 5-build save limit, and the add-to-cart card is hidden until we actually stock components. 445 passing.

"Deadly accurate" isn't a slogan — it's a test suite. This release is the product audit turned into shipped fixes, the day it was written.

BUG · fixedOne library everywhere — the builder and the tool pages now share the same catalogs

• The bug a tester caught: the arrow and bow pickers on My Arrows, My Bows, and the other standalone tool pages were missing models that the main builder had — and vice versa. Root cause: two copies of the component libraries, each edited separately for weeks.
• The fix: one canonical library now feeds everything, and an automated check blocks any future release where the copies drift apart. This can't quietly happen again.
• The merge made both sides better. The builder gains verified 2026 bows it was missing — Mathews ARC 30/34, Elite Varos/Artus/Terrain, PSE Mach 30/33 DS and Sicario FDS 33/35, Mission, APA, Truth Archery, Xpedition XLite — plus 65 nocks, inserts, points, and broadheads, sound-test data on broadheads, and a re-verification of every shaft's GPI table against manufacturer charts. The tool pages gain the newest animal vitals data and sight catalog.
• Removed phantom bows that never existed (Elite "Veros", non-DS 2026 Mach 30/33, a misnamed Sicario) — every 2026 entry is now verified against the manufacturer's published specs. 425 passing.

Thanks to the tester who noticed Victory models missing from the arrow choices. The library data is the product — drift between copies is a bug like any other, and now it's a tested one.

Catch-up — saves that survive, and a subscription tier that sticks

• v164 — saves never lost. Your saved builds and setups now also persist in the storefront page itself, not just inside the builder frame — so browsers that clear third-party storage can no longer eat your work.
• v163 — tier bridge root-cause fix. Signed-in Hunter/Pro subscribers occasionally loaded at the wrong tier; the page and the builder now actively handshake until the right tier lands. Also stopped asking subscribers for their email — we already know who you are.

Your subscription follows you to the standalone tools

• The dedicated tool pages — Tune Wizard, Wind-Drift, Build Review, Range Log, My Bows, My Arrows, Spine — now recognize your Hunter or Pro subscription the same way the main Forge does. Sign in and a paid plan unlocks the gated tools on those pages too; everyone else stays on the free set. Closes a gap where those pages didn't yet read your tier.

Subscriptions are live — and the paywall actually works

• Subscribe for real. Hunter and Pro now check out as true recurring subscriptions — monthly or annual — and a paid plan unlocks your tier inside The Forge automatically. Sign in and the tools you paid for are just there.
• Free is the way in. No trial gimmick: the Free tier is genuinely free forever, no card required. Fixed a gating bug where the builder could open at the wrong tier, and removed every "7-day trial" mention now that the Free tier is how you try it.

The honest version of a launch: the free tools stay free, the paid tools are paid, and what you subscribe to is exactly what unlocks. Nothing more, nothing less.

Subscriptions — corrected, consistent pricing everywhere

• The upgrade prompts inside The Forge now show the real numbers — Hunter $6/mo or $60/yr and Pro $12/mo or $120/yr (two months free on annual). Cleaned up an older modal that still quoted retired figures, and folded the separate Competition tier into Hunter (same price, same tools — target archers are welcome on Hunter). 337 passing.

Guided tour — a 60-second walkthrough of every tool

• New here? The Forge now greets you with a quick guided tour that steps through each tool — build, spine gauge, Penetration Index, sight tape, Range Log, Your Number, and the ethical-shot verdict — switching tabs as it goes so you see each one. Reopen it anytime with the Tour button up top.

Ethical-shot verdict — one go/no-go from accuracy × penetration

• On the Kill-Zone card your two signals combine into a single GREEN / AMBER / RED banner — GREEN only when you'll hit the vitals from there AND the arrow punches through. Either one failing vetoes the shot. 337 passing.

Accuracy puts it in the vitals; penetration punches it through. This is the line where both have to be true before you draw.

Penetration Index — the "deadly" half, paired with Your Number

• A 0–100 lethality grade on the build view with an Ashby factor scorecard — momentum, broadhead mechanical advantage, FOC, shaft diameter, structural integrity — plus the heaviest game class cleared and your weakest link. 330 passing.

Your Number — honest confidence at range, from your own groups

• v153 Range Log → Your Number turns your logged groups into a first-shot kill probability on any animal + a 90% ethical / 75% stretch max range (Hunter + Pro, unlocks at 20 plotted shots). v154 ranging-error toggle, long-range calibration from multi-distance groups, elliptical vitals across all 57 animals. v155 BUG measured launch speed in every mode (was a fixed 285 fps placeholder — a tester catch) + a "path to a longer number" coach. v156 crosswind dispersion folded into the kill probability. 324 passing.

Kill-Zone target, spine heat gauge, sight tape, and a corrected 2026 Mathews lineup

• v145 Kill-Zone ranging forgiveness + Spine Gauge stiff→perfect→weak (Hunter+); BUG removed non-existent "ARC 29.5/33", real lineup is ARC 30 + 34. v146–150 Mathews Lift RS/XD + UV Slider pin · PDF footer email fix · Shot Solver opened to Hunter · bespoke silhouettes · CTI grouped with the front assembly. v151 compact sight-tape numbers + full-color printing; BUG removed non-existent UltraView "Hindsight/UV Sniper" (real = UV3, UV3 SE). v152 brighter fluorescent tape colors.

Spine What-If — cycle the standard spine ladder + see which lands you in window

Field-tested via Jesse's PSE Mach 33 DS / K2 Sawtooth 4mm setup — spine 300 IN RANGE at his effective 83.75 lb, matching his real-world tune. 196 passing.

Account sync · Phase 1 — sign in, sync the Range Log across devices

CRITICAL BUG · fixedCross-origin parent.location read crashed the entire bundle

Lesson locked in memory: cross-origin parent access in a Shopify iframe always throws — wrap with try/catch or use postMessage only.

Share engine + Jesse's pro-tester fixes (spine parity, cut-length guardrail, Messenger-hijack fix)

Earlier sprint — Range Log, photo target + CV detection, Compare, sight tape, Methods, OG image, FOC Hybrid Model

v134 CV arrow-hole auto-detect · v133 photo-anchored logging · v132 Range Log Advanced · v131 Compare Calculators · v130 shot-by-shot plotting · v129 mobile + safety disclaimers · v128 Pin Layout & Caliper · v127 new emblem · v126 OG image v2 · v125 SVGDiagram aerospace · v124/v123 Field Notes · v122 string-stopper card · v121 WindDrift CFD · v120 Methods rewrite · v119 sight-axis + French tune · v117 Methods page · v115 Sparrow Hybrid FOC (reframe of v111) · v114 arrow-prep step · v113 nock-fit card · v111 trajectory + Ashby EFOC (partially reversed in v115) · v110 Wizard save + Express guard · v109 density-altitude unit fix.