A reflected-light meter does one thing: it renders whatever it reads as middle grey. That target is Zone V — roughly 18% reflectance — and every meter, in-camera or handheld, is calibrated to it. The arithmetic is clean, because one zone equals one stop equals one EV. So if you point the meter at a snowfield and obey it, the meter makes the snow grey: it has no idea the field is white, and it lands the frame about two stops dark. A backlit face is the mirror image — the bright rim and sky drag the reading up, and the shadowed face falls two to three stops under. Bracketing is the blunt hedge against this miscalibration. The two decisions that define a useful bracket are the increment between frames and the total spread, and both follow from knowing how far the meter has been fooled and how much error the emulsion will forgive.

The metering error, made numeric

Take a sunlit snow scene that meters at EV 15. Expose at that reading and the meter has placed the snow on Zone V; the white you wanted on Zone VII or VIII has been pulled down two to three stops, and the print comes back muddy and grey. The correction is to open up: add two stops to put the snow back on Zone VII. If you do not trust the correction, you bracket around it. A five-frame, full-stop bracket centred on the metered EV 15 records EV 13, 14, 15, 16 and 17 — spanning two stops either side. Spread is just (frames − 1) × increment: five frames at one stop span four stops total (±2); five frames at one-third stop span only four-thirds of a stop (±2/3). One-third is the finest step most bodies offer, because the ASA/ISO speed scale itself is subdivided in third-stops — 400, 500, 640, 800 — and a full stop is a single doubling or halving, one EV.

Expose for the shadows, develop for the highlights

Ansel Adams and Fred Archer’s working rule, set out in The Negative, makes much exposure bracketing unnecessary for the careful worker. Exposure fixes the shadows: you meter the darkest area in which you still want texture and place it on Zone III, two stops below the meter’s middle grey. Development then fixes the highlights, almost independently. Pulling development — N−1 or N−2 — lowers highlight density and compresses a contrasty scene, while N+1 expands a flat one, and both move the upper values with little effect on the shadows already anchored by exposure.

For sheet film, where every frame is developed on its own, this means the right tool is often a development bracket rather than an exposure one: expose three identical shadow placements and process them N, N−1 and N−2 to find the contraction the scene needs. The Zone System does not abolish exposure bracketing, but it explains why a Zone System photographer rarely reaches for it.

Latitude is film-specific — weight the bracket to the emulsion

There is no universal “one stop under, three over.” Latitude belongs to the emulsion. Ilford HP5 Plus is nominally ISO 400/27° and pushes cleanly to EI 3200/36°; in practice it tolerates about one stop of underexposure and two to three stops of over, its highlights extremely hard to block up. So a coarse, overexposure-weighted bracket suits it. Kodak Portra 400, by contrast, holds detail from roughly two to three stops under to five or six stops over — skewed so far toward overexposure that the same bracket placed symmetrically would waste half its frames on the wrong side. The asymmetry of your bracket should match the curve of the stock in the camera, not a remembered rule of thumb.

The practical consequence: most cameras bracket symmetrically about the metered value. To skew a negative bracket toward overexposure, dial in +1 stop of exposure compensation first, then let the AEB sequence run around that shifted centre. And when you bracket manually, change the shutter speed, not the aperture — depth of field stays put while exposure moves. At f/8, a five-frame full-stop bracket from a metered 1/250 s reads out as 1/1000, 1/500, 1/250, 1/125 and 1/60 s.

Reversal film inverts all of this. A slide has perhaps half a stop to one stop of latitude, because the dyes ride a steep characteristic curve and there is no printing stage to recover a misplaced tone — what the film records is the finished image. The standard transparency bracket, as Evident Scientific sets out in Fundamentals of Film Exposure, is therefore one-third-stop increments, two or three stops either side of the metered value. With Fujifilm Velvia 50, Provia 100F or Kodak Ektachrome E100, the fine increment is not caution but necessity.

Difficult light is where reciprocity ambushes you

Dim interiors, dusk and moonlight are the textbook “difficult light,” and there the meter is wrong before you bracket at all. Below about one second, film suffers low-intensity reciprocity failure: it records less density than the meter predicts, so the metered time is already too short. Ilford models the corrected time as Tc = Tm^P. For HP5 Plus, P = 1.31, so a metered 10 seconds actually needs 10^1.31 ≈ 20.4 seconds — call it 20. Exposures of one second or shorter need no correction at all.

This is film-dependent, and it changes how a long-exposure bracket is built. Fujifilm Neopan 100 Acros II needs no compensation under 120 seconds and only +1/2 stop from 120 to 1000 — which is why it is the standard night-and-long-exposure stock. Kodak Tri-X 400 fails harder, modelled near P ≈ 1.5; T-Max 400 is comparatively mild, around +1/3 stop at a metered one second. The rule is to compute the reciprocity-corrected time first and bracket around that, weighting further toward more exposure — never around the raw meter reading, which is the one value you already know to be too low.

The analogue answer to high dynamic range

When a scene’s range exceeds what one negative can hold, the digital answer is to merge a bracket into a tone-mapped composite. The slow-way answer is contrast control, and it usually needs only a single negative. Compensating and two-bath developers exhaust in the highlights while continuing to build the shadows, taming a contrasty scene at the development stage; N−1 or N−2 contraction does the same, as above. Then, in the darkroom, split-grade printing on variable-contrast paper such as Ilford Multigrade lets you expose shadows and highlights at different paper grades from the one frame — a low-contrast filtration for the highlight detail, a high-contrast pass for the shadows. Adams devotes much of The Print to exactly this division of labour. Done well, it recovers the same range a blended bracket chases, from a negative you can also still contact-print and file. The bracket, where you keep one, returns to its older and simpler role: insurance, several alternatives around a reading you were not sure of, from which you choose one and set the rest aside.