Every reflected-light meter built into a camera shares one assumption: that the area it reads should reproduce as a mid-tone. The meter measures reflected luminance and recommends an exposure that renders that average as middle grey. The pattern it uses to collect that average decides which parts of the frame steer the result, and therefore where the meter is reliable and where it is fooled. Knowing the pattern is half the battle; knowing the calibration constant behind it is the other half.

Why 12.5 percent, not 18

You will read everywhere that a meter is calibrated to 18 percent grey, the value of a Kodak test card. In practice it is not. A reflected-light meter solves for exposure using a calibration constant K, and the manufacturers split into two camps: Canon, Nikon and Sekonic build to K = 12.5, while Pentax, Kenko and the old Minolta line use K = 14. ISO 2720:1974 permits any K from 10.6 to 13.4 when luminance is measured in candela per square metre, so both conventions sit at or near the edge of the same standard.

The practical consequence is small but real. Point a K = 12.5 body at a true 18 percent grey card and it will not place the card on middle grey; it renders a slightly lighter tone as the mid-point, so the card reads about one-sixth of a stop differently from a K = 14 body. That gap between the two constants is the whole of the disagreement, since the same card seen through either calibration differs only by the ratio of the two values. If you cross-check with an incident reading, the two agree only on a surface whose reflectance is piK/C. With Sekonic’s K = 12.5 and incident constant C = 250 that is pi12.5/250 = 0.157, about 15.7 percent; using K = 14 gives about 17.6 percent, which is why a Pentax meter sits closer to the nominal card.

None of this matters much on HP5 Plus or Tri-X, whose latitude swallows a third of a stop without complaint. It matters on transparency film, where a third of a stop is visible, and it matters whenever you are placing a shadow deliberately and need to know exactly where your reading lands.

How center-weighted metering averages a scene

Center-weighted metering reads the whole frame but biases the result toward the middle. The weighting is not folklore; it is a published number. Nikon assigns 75 percent of the weight to a central circle and spreads the remaining 25 percent across the rest of the frame. On a body such as the D850 that circle defaults to 12 mm diameter for non-CPU lenses and is user-selectable to 8, 15 or 20 mm. The logic is statistical: in conventional composition the principal subject sits near the centre, so weighting that region gives a usable exposure across many ordinary scenes without any subject analysis.

The pattern fails wherever the bright or dark element does not coincide with the subject. A backlit portrait drives the average up and underexposes the face; a dark subject against a light wall does the reverse. Because the weighting is fixed, the meter cannot tell a deliberately high-key frame from an overexposed one, and snow, sand and white walls all read as mid-tone grey.

The snow correction nobody dials in

The white-surface failure is the most common and the easiest to fix. A meter pointed at snow drives that dominant white toward middle grey, which is why uncorrected snow comes back a dingy grey rather than white. The remedy is positive exposure compensation, scaled by how much of the frame the bright surface fills. A scene that is roughly 80 percent snow wants about +1.7 EV; a mixed landscape with a darker foreground and sky in the frame often needs only +1 EV. The working range is +1.5 to +2 EV for a frame dominated by snow or sand. The same logic runs in reverse for a dark subject filling the frame, which needs negative compensation to keep it dark.

Matrix metering, and what the numbers earned

Matrix metering divides the frame into discrete zones, compares their brightness relationships and selects an exposure by matching the pattern against a stored reference. Nikon introduced the approach in the Nikon FA in 1983 as Automatic Multi-Pattern metering, reading five segments: a central zone and four outer quadrants. (AMP was originally intended for the FE2 but was not production-ready in time.) The selection algorithm was empirical: Nippon Kogaku stated the program “was written after the visual assessment of nearly 100,000 photographs,” so a high reading at the top of the frame is interpreted as sky and discounted rather than averaged in.

The claim that matrix metering is opaque but better is one Nikon backed with figures: AMP delivered a good exposure 90 to 95 percent of the time against 85 to 90 percent for centre-weighting. Modern matrix systems extend this to hundreds or thousands of zones and add autofocus point, subject distance and colour. The weighting logic stays proprietary, so the result of a given frame cannot be predicted exactly, which is precisely why a deliberate worker reaches for a spot meter when the frame matters.

Overriding the pattern with a spot reading

Both averaging patterns produce a mid-tone, which is wrong whenever the subject is meant to be light or dark. The precise correction is a spot reading placed on the Zone System scale. A spot meter has a 1-degree acceptance angle; the canonical instrument is the Pentax Digital Spotmeter, introduced in 1977 and used by Ansel Adams in his later years, with the Sekonic L-758 as a modern equivalent. Many workers mark the meter’s dial with zone stickers for direct placement.

Any reflected reading renders the metered area as Zone V, middle grey. The zones run from 0 (black) to X (white), one stop apart, with V at the centre. To place a textured shadow on Zone III you stop down two stops from the indicated reading, because Zone III sits two stops below Zone V. Worked through: spot-meter a shadow with detail you want to keep and the meter reads, say, EV 9; that would render the shadow as middle grey. Set your exposure for EV 11 instead (two stops less light) and the shadow falls on Zone III, dark but textured, with the rest of the scale arranging itself above it. This is the operational core of the Zone System as Adams sets it out in the chapter on the zone system in The Negative (1981).

The alternative is to sidestep reflectance altogether. An incident meter uses constant C, on Sekonic about 250, to measure the light falling on the subject rather than the light it reflects, so subject tone never enters the calculation. A grey-card reading does the same trick from the camera position by substituting a known mid-tone for the unknown subject. Close the loop with the opening section, though: read an 18 percent card on a K = 12.5 body and the answer lands slightly off the card’s nominal value, which is why even the substitution method rewards knowing your meter’s constant.