Every decision about exposure and development eventually resolves into a single graph. The characteristic curve, also called the H&D curve, plots the optical density produced in a developed negative against the logarithm of the exposure that produced it. Ferdinand Hurter (1844–1898), a Swiss-born industrial chemist, and Vero Charles Driffield (1848–1915), an English engineer, published it in their 1890 paper Photo-Chemical Investigations and a New Method of Determination of the Sensitiveness of Photographic Plates in the Journal of the Society of Chemical Industry. That title is worth reading slowly: the curve and the first rational film-speed system arrived in the same paper, because once you can plot density against exposure you can also define where on that plot a film’s working speed lives. Reading the curve correctly explains why shadows lose separation when underexposed, why highlights block up, and why Tri-X 400 and T-Max 400 render the same scene so differently despite sharing an ISO.

Density Against Log Exposure

The horizontal axis is log exposure (log H), measured in lux-seconds; the vertical axis is density, the base-ten logarithm of the negative’s opacity. Both axes are logarithmic because the eye, the emulsion, and the exposure scale itself behave geometrically rather than linearly. One unit on the log H axis is a factor of ten in exposure. Since each stop doubles exposure and log10(2) = 0.301, one log-H unit is 1 / 0.301 ≈ 3.32 stops, and the inverse is the figure you actually meter to: 1 stop = 0.30 log H. A typical seven-stop subject brightness range therefore spans about 7 × 0.30 = 2.1 log-H units across the curve.

A useful negative does not begin at zero density. Even unexposed film, once developed, carries a small density from its grey base tint and from chemical fog. This baseline is base-plus-fog, or D-min, and every meaningful tone is measured as density above it. For a modern panchromatic film D-min typically falls around 0.18–0.25; Kodak’s sensitometry workbook uses 0.18 for its sample emulsion, and Adams assumed 0.10 for the idealised dye-free case. The anti-halation backing clears during development, so it adds nothing to the final D-min. The curve as a whole takes the shape of an elongated, tilted S: a slow start, a steep middle, a flattening top.

The Three Regions

The lower bend is the toe. Here density rises only gradually with exposure, so small differences in shadow exposure produce small differences in density. Tones placed deep in the toe are compressed and approach base-plus-fog, which is why severe underexposure erases shadow separation rather than simply darkening it.

Above the toe lies the straight-line section, where density increases in near-constant proportion to log exposure. The slope of this region is gamma, and gamma alone — it ignores the toe entirely. A steep slope stretches a given range of exposures across a wide range of densities (high contrast); a shallow slope compresses them (low contrast). Gamma is governed largely by development.

The upper bend is the shoulder, where each increment of exposure yields less additional density until the curve flattens at maximum density, D-max. Highlights driven into the shoulder are compressed toward a common tone, the negative equivalent of blown highlights.

Gamma Is Not the Contrast You Develop To

This is the distinction most curve diagrams skip. Gamma measures only the straight line, but manufacturers do not develop to a target gamma — they develop to an average gradient, which folds the toe back in. Kodak quotes Contrast Index (CI): the slope of a line drawn between two curve points 2.0 log-E apart, located by a marked straightedge whose zero rests on the D-min line so the lower point falls on the toe. Ilford quotes average gradient G-bar, measured over 1.50 log-H units from 0.10 above base-plus-fog. Both deliberately include the toe.

The consequence is the crux of the whole subject: two films can share an identical straight-line gamma and still print differently, because their toes differ in shape. A long, gentle toe rolls shadows in slowly; a short toe snaps from threshold to full slope. The average gradient captures that, gamma does not. That is why a datasheet shows you contrast-time curves keyed to CI or G-bar rather than to gamma.

Reading One Curve, Number by Number

Work Kodak’s own sample to make this concrete. Gamma first: their figure rises from density 0.64 at log H 1.5 to density 1.58 at log H 3.0, so

γ = (1.58 − 0.64) / (3.0 − 1.5) = 0.94 / 1.5 = 0.63.

Now the average gradient from the same workbook, which starts on the toe. With D-min 0.18, take point A at density 0.28 (log H 0.9), then count 1.30 log-E units across to point B at density 1.08. The rise is 1.08 − 0.28 = 0.80 over 1.30 log-E:

G-bar = 0.80 / 1.30 ≈ 0.62.

That 0.62 is not a coincidence. It is exactly the contrast the ISO 6 speed standard demands, which is the next section. Once you can run those two subtractions, you can read contrast off any datasheet curve without trusting the printed label.

Where Film Speed Lives

ISO 6:1993, the standard for black-and-white negative film, fixes the speed point at the exposure giving a density of 0.10 above base-plus-fog, low on the toe where the first usable shadow texture appears — the same place Hurter and Driffield first looked for a rational speed criterion. Crucially, the standard also fixes the contrast at which the measurement is made: the film must be developed so that a second point, 1.30 log-E units above the speed point, reaches a density 0.80 greater than the speed-point density. That 0.80 rise over 1.30 log-E is itself an average gradient of 0.80 / 1.30 ≈ 0.62 — so the standard bakes a specific development contrast into the speed number, which is why the worked example above lands on the same figure. Arithmetic speed then follows from S = 0.80 / Hm, where Hm is the exposure in lux-seconds at the speed point, rounded to the nearest standard value.

In practice the contrast target lives on the datasheet as a development time. Ilford’s published HP5 Plus characteristic curve is for 6½ minutes at 20°C in Ilfotec HC (1+31) stock, with intermittent agitation; the same datasheet’s table gives EI 400 times of 7½ minutes in ID-11 stock or 13 minutes in ID-11 at 1+1 dilution — times Ilford describes as producing “negatives of average contrast suitable for printing in all enlargers” over a recommended EI range of 400/27° to 3200/36°. Move the temperature and the time tracks it: Ilford’s own rule gives 6 min at 20°C ≈ 4½ min at 23°C ≈ 9 min at 16°C. Longer time, higher temperature, or a more active dilution raises the average gradient; pull development back and it falls. That is the practical lever behind the abstract word “gamma.”

Same ISO, Different Curve

Now the intro’s claim, demonstrated. Kodak Tri-X 400 has a long toe and a slight shoulder. The long toe rolls shadows in gently, and the shoulder self-compresses highlights, so the film forgives overexposure and a contrasty light gracefully — part of why it became the reportage standard. Kodak T-Max 400 (TMY-2) is a short-toe, near-straight emulsion with essentially no shoulder. It climbs to D-max almost in a line, giving cleaner shadow separation and crisper highlight gradation, but it punishes shadow underexposure because there is little gentle toe to fall into and little shoulder to catch blown highlights. Both are nominally ISO 400. Meter them the same and they record the same scene differently — not because their speed differs, but because the shape of the curve between toe and shoulder differs.

Metering onto the Curve

This is where the curve meets the Zone System. Ansel Adams’ density anchors (The Negative, 1968) map directly onto it: he arbitrarily assumes a base-plus-fog of 0.10, places Zone I at ≈ 0.10 above base-plus-fog — the first usable shadow texture, coinciding with the ISO speed point — and a correctly exposed and developed Zone V at density 1.10 above base-plus-fog (total density 1.20). Placing a shadow on Zone III means seating it just above the toe, two stops up from Zone I, where gradation has opened out. Exposure positions the scene along the log H axis: everything below the toe collapses toward base-plus-fog, everything above the shoulder merges toward D-max, and the working portion between them must hold the subject’s range. A seven-stop scene is that 2.1 log-H units from earlier — it has to land between toe and shoulder, or you lose an end.

Development then rotates that section about the speed point. The reason the toe stays comparatively anchored is mechanical: the developer reduces exposed silver-halide grains starting from their latent-image centres, and heavily exposed highlight grains carry far more of those centres than threshold shadow grains. With extended development the highlight grains gain density fastest while the near-threshold shadow grains barely move, so the upper curve swings up and the toe holds — exactly the family of contrast-time curves a datasheet prints. Read this way, the characteristic curve is less a specification than a map of every exposure and development choice a negative can hold.