Every metered exposure carries some error: a misread highlight, a backlit subject, a guess in changing light. What separates one medium from another is not whether error occurs but how much error can occur before the image fails. This tolerance is exposure latitude, and black and white negative film and digital sensors handle it in almost opposite ways. Understanding the difference clarifies why one medium is forgiving in the highlights and the other in the shadows, and which way to bias your meter for each.

Latitude Is Not Dynamic Range

The two terms are routinely confused. Dynamic range is the total span of luminance a medium can record, from the deepest shadow that registers above noise to the brightest highlight before saturation. It is a fixed property of the material. Latitude is the margin of exposure error a scene permits while still landing within that range. A scene of low inherent contrast leaves room to misexpose by several stops and still fit; a scene that already spans the full dynamic range leaves none.

Latitude therefore depends on both the medium and the subject — and, in long exposures, on reciprocity failure, which quietly shrinks it. Below about 1/2 s, FP4 Plus and HP5 Plus need no correction down to 1/10000 s. Beyond half a second the film loses sensitivity and Ilford gives exact corrections: for FP4 Plus the adjusted time is Ta = Tm^1.26, for HP5 Plus Ta = Tm^1.31, where Tm is the metered time. A two-second metered reading on HP5 becomes about 2.5 s on the timer; an eight-second reading becomes roughly 15 s. Miss that and the shadows, already on the steep part of the curve, fall away first — latitude collapses regardless of how forgiving the emulsion is in normal light.

Reading the Characteristic Curve in Numbers

The behaviour of black and white negative film is governed by its characteristic curve, which plots developed density (D) against the logarithm of relative exposure (log H). The horizontal axis is logarithmic for a reason: one stop equals 0.30 in log-E, because doubling exposure adds log₁₀(2) ≈ 0.30. The curve bends out of a toe at low exposures, climbs through a long, nearly straight line, and rolls into a shoulder at high exposures, but the transitions are gradual rather than sharp.

The slope of that straight line is gamma, and the practical contrast metric for printing is the Contrast Index (the average gradient over the useful range). For a diffuse enlarger you aim for a CI of about 0.55–0.62 to print on grade 2; a condenser head, which raises apparent contrast, calls for a softer negative nearer 0.50. Development controls exactly this. FP4 Plus at EI 125/22 in ID-11 stock at 20°C runs 8½ minutes for negatives of average contrast; the same film in ID-11 diluted 1+3 takes 20 minutes and yields a flatter, lower-gamma curve. That is what “density rises in proportion to exposure” means quantitatively: a measurable gradient you set with dilution, time and temperature.

Why Negative Film Forgives Overexposure

Because additional exposure simply pushes tones further up a still-rising slope, highlights compress slowly instead of vanishing. This gives negative film generous and asymmetric latitude. Overexposure adds density and grain but preserves separation; underexposure starves the shadows, which fall onto the toe and lose detail first.

The HARMAN/Ilford FP4 Plus datasheet (Nov 2018) puts numbers on it. The film is rated ISO 125/22, measured in ID-11 at 20°C with intermittent agitation, with a recommended exposure-index range of EI 50/18 to EI 200/24. The datasheet states it “will give usable results even if it is overexposed by as much as six stops, or underexposed by two stops” — six over, two under. HP5 Plus, ISO 400/27, behaves differently: its documented latitude is a push range, not a shoulder claim. The datasheet shows usable, high-quality prints from box speed EI 400 up to EI 3200/36, with extended development in ILFOTEC DD-X or MICROPHEN (stock) for maximum speed — DD-X (1+4) is 9 minutes at EI 400, with longer tabulated times for each push.

A Worked Zone System Placement

The asymmetry has a name and an operating rule. Ansel Adams set it out in The Negative (1981): expose for the shadows, develop for the highlights. A reflected-light meter, by calibration (ISO 2720), renders whatever it reads as a mid-tone — Zone V, conventionally the 18% grey card. Each zone is one stop, so to place a tone elsewhere you offset from the meter’s reading.

Work an example with FP4 Plus. Spot-meter a shadowed stone wall where you want texture and it reads 1/60 at f/8. Trust that and the meter drops the wall on Zone V — too light, texture flattened. Texture with shadow weight wants Zone III, two stops down, so you close down two stops to 1/250 at f/8. Now check the brightest sunlit render on the wall: if it falls three to four stops above the shadow, it lands on Zone VI–VII and prints with bright detail. To pull a hot highlight from Zone IX back to VIII you give N-1 development; to lift a flat scene you give N+1.

Here is the latitude in action. Suppose you misjudge and over-expose that frame by three stops. On FP4, with six stops of overexposure headroom, you are still comfortably inside the straight line — the negative is dense and grainier but every tone holds. Make the same three-stop error on a sensor metered near its ceiling and the highlights are already clipped: gone, with no shoulder to soften the loss. The practical lever behind film’s tolerance is simply that you can rate FP4 at EI 64–100, deliberately over-exposing to seat the shadows higher up the curve.

Why Sensors Clip, and Why You Expose to the Right

A digital sensor responds in the opposite manner. Each photosite is a well that accumulates charge linearly with light until it saturates. There is no shoulder: once a well fills, every pixel above that point records the same maximum value and the detail is gone permanently. The result is an abrupt clipping ceiling.

The linearity is the whole story, and it is quantifiable. A 14-bit converter has 16,384 code values, but because the encoding is linear the brightest stop of the captured range claims 8,192 of them, the next stop 4,096, then 2,048, 1,024, 512 — halving every stop down the scale. By the fifth stop below saturation only about 512 levels describe the tones; the darkest stops carry a handful. That, plus a fixed read-noise floor, is why lifting shadows in processing looks gritty and posterised: there is almost no data down there to lift.

This inverts film’s strategy. A modern full-frame sensor records roughly 13–15 stops of dynamic range at base ISO — the Nikon D850 and Z7 measure near 14.6–14.8 EV of landscape (engineering) dynamic range at base ISO 64 on DxOMark, though Bill Claff’s stricter Photographic Dynamic Range metric puts the same sensors nearer 11 stops — a span broadly comparable to film, but its usable latitude lives in the shadows, not the highlights. So you expose to the right (ETTR): place the brightest important, non-specular highlight just short of clipping. Shifting the whole signal up the scale lifts it above the read-noise floor and buys roughly one to two stops of effective dynamic range in the deep shadows.

The two rules are mirror images. Film: expose for the shadows, develop for the highlights — rate FP4 at EI 64–100 and place shadows on Zone III. Digital: expose for the highlights, recover the shadows — meter to just short of clipping. Each medium is exposed to protect the end of the tonal scale it cannot recover, and the meter is biased the opposite way for each.

Sources: HARMAN/Ilford FP4 Plus and HP5 Plus Technical Information datasheets, Nov 2018; Ansel Adams, The Negative (1981); DxOMark sensor measurements (dxomark.com) and Bill Claff, Photons to Photos (photonstophotos.net).