Shape Is the Product

Pick up a kernel of H4831 and a pinch of Titegroup. One is a stubby perforated cylinder you could almost thread on a wire. The other is a flake the size of a coarse-ground pepper fleck. Both are nitrocellulose. Both burn. The difference between them is not mainly chemistry — it is geometry and dimension, and that difference is the entire product.

A propellant is not a fuel. It is a machine for producing a specific pressure-versus-time curve inside a barrel. The chemistry sets how much energy is available. The shape decides when it arrives.

The problem shape solves

When a grain of powder ignites, it burns from every exposed surface inward, at a rate governed by pressure. The flame front eats into the solid at more or less the same speed everywhere it touches, which means the rate of gas production at any instant is proportional to total burning surface area at that instant.

And here is the difficulty. As a solid grain burns, it gets smaller. Smaller means less surface. Less surface means less gas. So the natural behavior of a burning solid is to produce its gas fastest at the very beginning — right when the bullet has not moved yet, the chamber volume is smallest, and a pressure spike is exactly what you do not want.

Meanwhile the bullet is accelerating down the bore, and the volume behind it is growing rapidly. To keep pushing, you need gas production to hold or even climb as the burn proceeds.

Nature gives you the opposite. Every shape in a modern propellant catalog is an attempt to fight that.

Degressive, neutral, progressive

Propellant geometries are classified by what happens to burning surface as the grain is consumed.

Degressive — surface decreases as it burns. A sphere is the purest case: surface goes as the square of the radius, so a sphere sheds surface fast. Solid cylinders and flakes are degressive too, though more gently.

Neutral — surface stays roughly constant. The classic solution is a single-perforation tube. As it burns, the outside shrinks and the hole in the middle grows. For a long tube those two effects cancel almost exactly: the outer circumference loses precisely what the inner circumference gains. The arithmetic is unusually clean, which is why the single-perf tube has been in continuous use since the nineteenth century.

Progressive — surface increases as it burns. Put seven perforations in the grain instead of one and you have seven growing inner surfaces set against one shrinking outer surface. The growth wins, and burning area climbs steadily through most of the burn.

The seven-perf grain has a catch. At some point the perforations burn into each other and into the outer wall, and the grain collapses into a set of small three-cornered fragments called slivers. Surface area falls off a cliff, which is why the progressive curve terminates abruptly rather than tapering. This is not an abstraction: in reduced-charge artillery firings, unburned slivers are found on the ground beyond the muzzle. You can pick the geometry lesson up off the grass.

Each profile suits a different job. A pistol cartridge with a short barrel and a light bullet needs its energy released almost instantly — degressive flake is right. A large artillery piece with a long barrel needs sustained push over a long travel — multi-perforation progressive is right. Rifle cartridges live in between.

Web thickness: the variable nobody names

Here is the thing most reloaders never get told. Handloaders talk about “grain size.” Propellant engineers talk about web — the thickness of solid material the flame front actually has to burn through before the grain is consumed. In a single-perf tube, the web is the wall thickness between the outside and the hole, not the length and not the outside diameter. For the seven-perf grain the geometry is regular enough that interior-ballistics texts give the web as a formula: (D − 3d) / 4, where D is the grain diameter and d the perforation diameter.

Two dials set burn rate within a chemical family: the web, and the deterrent coating (covered below). A thick web takes longer to burn through, which means a slower powder. A thin web burns through quickly, which means a fast one.

This is why powders within a family form a ladder. H4895, H4350, and H4831 are close chemical relatives — single-base extruded powders from the same Australian plant, each better than 90 percent nitrocellulose. What separates them, and what makes one right for .223 and another right for magnum rifle cases, is mostly web and deterrent level. Same recipe, different wall.

It is also why the total burn is best understood as a race: the powder must finish burning at roughly the point in bore travel where the pressure curve and the bullet's position line up favorably. Too thin a web and it is all over before the bullet has gone far — high peak, poor efficiency. Too thick and unburned powder exits the muzzle as flash and wasted mass.

How extruded powder is made

This is the older process and still the one used for most rifle propellant. It runs one to two weeks end to end.

Nitration. Purified cellulose — cotton linters, or increasingly wood pulp — is treated with a mixture of nitric and sulfuric acid. The sulfuric acid is not a reactant; it absorbs the water the reaction produces and drives it forward. How much nitrogen ends up bonded to the cellulose determines how energetic the result is, and the grades have names: pyrocellulose at about 12.6 percent nitrogen dissolves in ether-alcohol; guncotton at 13.35 percent and up does not; military-grade nitrocellulose is a deliberate blend of the two, specified around 13.15 to 13.25 percent. Blending is built into this product at the molecular level before a single grain exists.

Purification. This is the step that killed people for two decades in the 1800s. Residual acid trapped in the fibers attacks the nitrocellulose and generates more acid, which accelerates the attack. The material is boiled, pulped to open up the fiber structure, and washed exhaustively. Frederick Abel's insight — pulp it, then wash it — dates to 1865, after a string of fatal factory explosions, and it is still the basis of the process. The stakes of getting stability wrong were written in steel: residual instability in early French smokeless powder is blamed for the magazine explosions that destroyed the battleships Iéna in 1907 and Liberté in 1911, at anchor, in harbor.

Dehydration. Nitrocellulose is stored and transported wet, because wet nitrocellulose is comparatively safe. Before use, the water is displaced with alcohol in a press rather than dried off with heat.

Colloiding. The dehydrated nitrocellulose goes into a heavy mixer with an ether-alcohol solvent and a stabilizer such as diphenylamine. This is Paul Vieille's 1884 discovery industrialized: nitrocellulose gelatinized into a colloid burns predictably, layer by layer, instead of erratically — the insight behind Poudre B, the first practical smokeless powder. Alfred Nobel added the second idea three years later: dissolve nitroglycerin into the colloid and you get Ballistite, the ancestor of every double-base powder. Modern double-base extruded powders still work Nobel's way, with nitroglycerin worked into the dough.

Blocking and screening. The dough is forced through fine screens to catch any foreign particle, then pressed into blocks. A hard inclusion in a propellant grain is a defect that will show up as a flyer or worse.

Extrusion. A hydraulic press forces the colloid through a die. The die opening sets the outside profile; pins suspended in the opening create the perforations. One pin gives a single-perf tube, seven pins give a seven-perf grain. The material emerges as a continuous cord.

Cutting. A rotary cutter chops the cord to length. Grain length relative to diameter matters — it affects how grains pack, how they flow, and how much of the burning surface comes from the ends.

Solvent removal and recovery. The solvent has to come out, slowly and evenly, or the grains crack or warp. Air drying is followed by water drying, and the ether and alcohol are recovered and reused. This is the slow step, and it is why extruded powder takes weeks rather than hours.

Finishing. Coating, glazing, screening, blending — covered below.

How ball powder is made

Fred Olsen's process, developed by 1933 at Western Cartridge Company, is a completely different animal, and understanding why it exists explains a lot about the modern market. It was invented as a salvage process: the U.S. had tons of First World War cannon powder slowly deteriorating in storage, and Olsen worked out how to dissolve it and re-form it into stable, usable propellant. The economics of that origin never left the product.

Lacquer. Nitrocellulose — virgin or reclaimed — is dissolved in ethyl acetate to make a syrupy lacquer. Stabilizer goes in here.

Emulsion. The lacquer is poured into water containing a protective colloid and agitated hard. It breaks into droplets, and the droplets are agitated until they reach the size wanted — the operator dials in the grain dimension with a stirrer rather than a die.

Distillation. The batch is heated under controlled conditions and the ethyl acetate is driven off and recovered. As the solvent leaves, each droplet hardens into a solid sphere. Dewater, dry, and screen to size.

Impregnation. For double-base powder, nitroglycerin is added after the spheres are formed, soaking into the existing grain rather than being mixed in from the start.

Rolling. Here is the clever part. Spheres are passed between rollers and flattened. Flattening a sphere reduces its web while increasing its surface, which makes it faster — and lets a manufacturer take one base sphere and turn it into several distinct powders by rolling to different degrees. Most “ball” powders on the shelf are actually flattened ball, and the label often says so.

Finishing. Coat, glaze, screen, blend.

The process is fundamentally cheaper and faster than extrusion — on the order of forty hours against one to two weeks — runs as batch chemistry rather than mechanical forming, and tolerates reclaimed nitrocellulose that would be unusable in an extrusion line. That economics is why spherical propellant dominates high-volume military small-arms production, and why St. Marks in Florida can turn out something on the order of 6,000 tons a year. It is also woven into handloading's own origin story: some of the first surplus powder Bruce Hodgdon sold civilians was spherical propellant pulled from .303 British military cartridges — sold as “BL type C,” a designation that survives on canisters today as BL-C(2).

Its limitation is web. Building a really thick web into a sphere is hard, which is why the slowest magnum rifle powders remain extruded.

Coatings, glazing, and the finish

Three things happen to a finished grain, and all three matter more than they sound.

Deterrent coating. The outer layer of the grain is treated with a slower-burning compound — dinitrotoluene, dibutyl phthalate, or a centralite. The grain then burns slowly at first and accelerates as the flame front reaches uncoated material underneath. This is progressive burning achieved chemically instead of geometrically, and it is how a manufacturer converts one base grain into a family of distinct products. Do not picture a trace dusting: on the safety sheet for a current extruded magnum rifle powder, the deterrent runs as high as one part in ten of the whole grain. It is a full ingredient, and it is why you should never assume two powders with similar-looking kernels behave alike.

Graphite glazing. Grains are tumbled with graphite. This does three jobs at once. Nitrocellulose is an electrical insulator, and dry powder flowing through a measure generates static that makes kernels cling, clump, and meter inconsistently — graphite bleeds that charge away. It improves flow generally. And it keeps grains from sticking to each other and to machinery. The black dust on your hands and in the bottom of the jug is doing real work.

Screening. Undersize and oversize kernels are removed. Consistency of grain dimension is consistency of burn rate.

The finish chemistry is still a live area of development, and not only for pressure curves. Flash suppressants — potassium salts, under one percent — are blended into many rifle powders. Decoppering additives that reduce jacket fouling are now a selling point: Hodgdon's CFE line takes its name from “Copper Fouling Eraser,” and Vihtavuori blends an anti-fouling agent across its range.

Blending, and why canister powder costs more

No propellant plant hits its ballistic target exactly on any given production run. Nitration varies. Solvent removal varies. Lots differ.

Military and commercial bulk powder is largely sold as made — the loading plant tests each lot and adjusts its charge weights to suit. At military scale the averaging is industrial: lots of ten tonnes and more are homogenized through towers of blending hoppers. That works fine when you are loading millions of rounds under laboratory supervision and re-qualifying with every lot.

Canister-grade powder is different. Production lots are tested and then blended together to hit a published burn-rate specification, so that the can you buy in 2026 behaves like the can you bought in 2019. That blending step is what makes published load data possible. Without it, every manual would need a lot number.

Three consequences worth carrying to the bench:

• Canister powder costs more than bulk for a reason, and the reason is a QC and blending operation, not marketing.
• Lot-to-lot variation still exists — it is controlled to a tolerance, not eliminated. Rechecking a load when you open a new lot is sound practice, not superstition.
• A powder sold as “bulk” or surplus may be chemically identical to a canister grade and still not match published data, because it never went through the blend.

What all of this means at the powder measure

Grain geometry does not stop mattering when the powder leaves the plant. It determines how the stuff behaves in a drop tube, and that shows up directly as charge weight variation.

Spherical and flattened ball meter beautifully. Spheres flow, pack to a consistent bulk density, and do not interlock. If you throw charges volumetrically at volume, this is the easiest class of powder you will handle.

Extruded stick is the difficult one, and the longer the grain relative to its diameter, the worse it gets. Long cylinders bridge across the drop tube opening, orient themselves differently from throw to throw, and get sheared by the measure. Cut kernels are a normal consequence, not a malfunction. Short-cut extruded powders behave noticeably better than long-kernel ones.

Flake splits down the middle. Small, dense flakes — Titegroup is the standard example — meter nearly as well as ball; that is much of why fast pistol powders are flaked. Large flakes are another matter: platy grains stack unpredictably and hold static, making coarse flake the class most sensitive to humidity and to whether the graphite glaze is doing its job.

None of this is a defect. It is the direct, physical consequence of shaping a grain for a pressure curve rather than for a powder measure. The geometry that makes H1000 right for a magnum case is the same geometry that makes it awkward to throw — and the geometry that makes Titegroup meter like water is the same geometry that makes it unforgiving of a double charge, because a light, fast, degressive flake fills so little of the case that a double charge still fits.

That last point is worth sitting with. The shapes that meter best are frequently the ones where a metering error is hardest to see and most dangerous. A case with twice the intended charge of a bulky extruded rifle powder overflows. A case with twice the intended charge of a dense, fast pistol powder looks entirely normal — which is precisely the failure a powder-level sensor exists to catch on a progressive press.

Process descriptions here are general to the industry; individual manufacturers vary in sequence, solvent system, and finishing.

Sources

• Weapons propulsion and grain geometry — U.S. Navy officer text, ch. 16 (FAS mirror)
• Naval Ordnance (1937), ch. III — interior ballistics, web formulas, sliver burning
• U.S. Army nitrocellulose manufacturing appendix — nitration grades and blending
• Ball propellant — process history and Olsen's salvage origins
• Poudre B — Vieille's colloided single-base powder and its early instability
• Thales/ADI extruded powder safety data sheet — composition of a current single-base rifle powder%20-%20407_default.pdf)
• Lee Precision — metering quality of powder types
• Guns & Ammo — “You Are Probably Shooting St. Marks Powder”

Elsewhere on the Desk

• The Propellant Squeeze — Gunpowder costs, supply, and industry direction — August 2026
• Who Actually Makes Your Powder — The dozen facilities behind every label on the shelf — and where each stands in 2026
• The Great Unbundling — How the powder industry took itself apart — and why it is putting itself back together
• The powder directory — 228 powders, verified against the makers' own documentation
• Prices & availability — what a pound costs in hand, sampled from retailers that permit it
• The merged burn-rate chart

Verify the charge, not just the recipe

A published load tells you what you meant to put in the case. It cannot tell you what went in on stroke 340 — a bridged measure, a double charge, a squib. Only measuring every round tells you that, and that measurement is what an RCM Laser Powder Check produces.