Every exposure meter answers the same question with one of two fundamentally different measurements. A reflected meter reads the light leaving the subject; an incident meter reads the light falling on it. The distinction is not academic. The two methods can disagree by two stops or more on the same scene, and understanding why explains one of the most common sources of mis-exposure in black-and-white work.

A Snowfield, Metered Two Ways

Take a sunlit snowfield on a roll of HP5 Plus rated at ISO 400. Bright front-lit sun is the Sunny 16 case, f/16 at 1/400s, about EV 15 at ISO 100, and that is your meter-independent sanity check before you trust anything the cell tells you.

Point an in-camera reflected meter at the snow and it reads the brilliant white field, then does the only thing it knows how to do: it renders that field as a middle tone. It hands you something like f/16 at 1/500s. Place an incident meter in the same sun, dome toward the camera, and it ignores the snow’s reflectance entirely; it reports the illumination, and gives you roughly f/8 at 1/500s. Two stops apart, on one scene, from two correct instruments.

Now read the result on the Zone scale. The reflected exposure puts the snow on Zone V, middle grey, dingy and underexposed by about two stops. The incident exposure, two stops more open, lifts that same snow to Zone VII, textured white with detail still on the straight part of HP5’s curve. To fix the reflected reading by hand you open +2 to +2.5 stops over the meter’s indication, since snow belongs on Zone VII or VIII, which is exactly the f/16-to-f/8 jump the incident meter handed you for free.

What Each Meter Actually Measures

A reflected meter, including every in-camera meter and every spot meter, samples luminance: the light bouncing off the scene toward the lens. Because that quantity depends on how reflective the subject is, the meter cannot know whether a dark reading comes from dim light or a dark surface. To produce a single exposure value it assumes the area it reads is of average reflectance and renders it as Zone V.

An incident meter sits at the subject position behind a translucent dome and measures illuminance: the light arriving at the subject regardless of what the subject later does with it. A white wall and a black coat under the same lamp return the same incident reading, because the meter never sees either surface. Since it sets exposure from the illuminance alone, reflectance then maps linearly onto the film’s curve: whatever is middle-grey lands on Zone V, a 90% white near Zone VII, a 3% black near Zone III, spanning the roughly seven-stop reflectance range of an ordinary scene without any correction from you.

Both methods are codified in ISO 2720:1974, the guide to product specification for general-purpose photoelectric exposure meters. Its constants are not derived from first principles; the standard fixes K and C by statistical analysis of viewer acceptability across many test photographs, so they are perceptual consensus values. The reflected constant K is recommended in the range 10.6 to 13.4; in practice Canon, Nikon and Sekonic use K = 12.5, while Minolta, Kenko and Pentax use K = 14. That gap is about 1/6 EV, so a Minolta or Pentax body exposes very slightly less than a Sekonic for the same scene. The incident constant C depends on the receptor: ISO 2720 allows 240 to 400 lux for a flat cosine-responding receptor (250 is common) and 320 to 540 lux for a hemispherical cardioid dome, with practical values from 320 (Minolta) to 340 (Sekonic).

Doing the Arithmetic on Middle Grey

The familiar “18% grey” shorthand is not the figure the standard actually agrees on. The reflectance at which a reflected reading and an incident reading land on the same exposure is pi x K / C. For K = 12.5 and a flat-receptor C = 250 that is pi x 12.5 / 250 = 0.157, about 15.7%, and a reflected meter on its own corresponds to roughly 12 to 12.5% reflectance, not 18.

This is why a grey card is not quite a free lunch. Twelve-percent grey is about half a stop lighter than 18% grey, so a meter reading taken off a true 18% card will underexpose by that half stop. Kodak prints the fix on its own Neutral Test Card R-27: meter the centre of the card from the camera direction, and “for subjects of normal reflectance, increase the indicated exposure by one half stop.” That card carries an 18% grey side and a 90% white side, with reference patches at 3% black and 90% white. Aiming matters too: Kodak’s instruction is to turn the card toward a point one third of the angle between the camera-to-subject axis and the main light, so for a light 30 degrees to the side and 45 degrees up you angle the card 10 degrees to the side and 15 degrees up.

Even an ordinary scene is not neutral on average. Ansel Adams notes in The Negative that an uncompensated on-axis reflected reading wants about +1/3 stop for roughly 85% of typical scenes, the same direction as the grey-card correction and for the same reason.

Naming the Hardware

The receptor is the part that does the thinking. On a Sekonic L-308X the hemispherical dome (cardioid response) integrates the whole lighting set-up over a hemisphere; swap it for the flat Lumidisc (cosine response) and you isolate a single source or read a flat copy plane, which is how you measure lighting ratios or evenness across artwork. Older meters do the same thing mechanically: the retractable Lumisphere on a Sekonic L-358 or L-608 extends for incident readings and retracts for cosine-corrected flat measurement.

Spot work is a third instrument. A Pentax Digital Spot Meter or a Sekonic with a 1-degree finder reads a narrow patch of luminance, which lets you place a tone deliberately. Spot a deep shadow where you still want texture, note the reading, then close two stops from the meter’s middle-grey indication: that shadow now sits on Zone III, the lowest zone that holds genuine detail on a film like FP4 Plus. The incident meter measures the light; the spot meter measures the scene. Knowing which question you are asking is the whole of the technique.

When Incident Metering Also Fails

Incident metering does not “sidestep the trap entirely”, because it answers a narrower question than it appears to. It measures the light on the subject, so it goes wrong whenever the picture is not about that light. A backlit or rim-lit subject reads from the front-facing dome as though the key light were doing the work, and the face goes dark; specular highlights and translucent subjects such as glass, smoke or a backlit leaf carry no surface reflectance for the dome to stand in for; and a distant landscape, a stage performer or anything you cannot physically reach denies you the one thing the method requires, a dome in the same light as the subject. There is also the deliberate case: when you want a low-key portrait or a snowfield rendered as menacing grey, a literally correct incident reading is the wrong answer, and the spot meter and the Zone scale are how you take the rendering back into your own hands.