A contrast filter is often treated as a fixed tool: a red filter darkens skies, a yellow filter holds them. In practice its effect depends as much on the light striking the scene as on the glass itself. A filter removes part of the spectrum, but the source decides how much of that part was present to begin with. The same filter on the same film can produce a markedly different tonal separation, and demand a different exposure increase, under tungsten light, open daylight, or the blue light of shade. The clearest proof is that a manufacturer’s own factor tables list two columns, daylight and tungsten, and for some filters the two numbers move in opposite directions.

The mechanism: where the light actually sits

A contrast filter works subtractively. It transmits its own colour and absorbs the complementary band, so a deep red filter passes long wavelengths and blocks blue and green. The tonal shift it produces depends on how much of the absorbed band the source actually emitted, and that is governed by colour temperature. Daylight for black-and-white work is conventionally balanced to about 5500 K (the CIE D55 standard), rich in blue and ultraviolet. Studio tungsten runs around 3200 K; household tungsten is cooler still, roughly 2700-2900 K; Ilford measures the tungsten speed of its ORTHO Plus film against a 2850 K source, the figure printed beside the datasheet’s wedge spectrogram. Open shade, lit by blue skylight rather than direct sun, climbs the other way to roughly 7000-10000 K and pushes the balance further toward short wavelengths.

The reason blue is so scarce under tungsten is the shape of the blackbody curve. A 3200 K emitter peaks near 905 nm, deep in the near-infrared: most of its energy is dumped into red and infrared and very little falls in the blue end of the visible band. A red filter under tungsten therefore discards almost nothing, because almost nothing it blocks was present, while a blue filter is left straining to pass a band the source barely emits. Under 5500 K daylight the situation reverses, with ample blue for a yellow or red filter to remove. The filter is constant; the spectrum it filters is not.

Reading factors as stops

A filter factor is a multiplier on exposure. Kodak states the rule plainly on the Tri-X datasheet: multiply the normal exposure time by the filter factor. The conversion to stops is logarithmic, and worth committing to memory: factor 2 is one stop, factor 4 is two stops, factor 8 is three stops, each doubling of the factor adding one more stop. So a factor of 6 is about 2.6 stops and a factor of 12 is about 3.6 stops.

A worked example on Kodak Professional Tri-X 400, an everyday panchromatic stock, shows the inversion the theory predicts. Meter the scene without the filter; suppose the camera reads f/8 at 1/125 s. With a No. 25 red filter in daylight, the factor is 8, so add three stops: open to f/8 at 1/15 s, or hold the time and open to roughly f/2.8. Move the same film and filter under tungsten and Tri-X’s factor for the No. 25 red drops to 5, about 2.3 stops, because the red filter is now wasting very little light. The No. 47 blue does the opposite: its factor is 6 in daylight (about 2.6 stops) but climbs to 12 under tungsten (about 3.6 stops, a full stop more correction) because the blue glass is fighting a source that has almost no blue to give. Same film, same two filters, and the required corrections cross over with the light.

Where the rule stops: the green band

“Factors always shift” is too strong. The shift is largest at the spectral extremes, the blue-passing and blue-blocking filters, and smallest through the middle. On Tri-X 400 the No. 58 green filter is factor 6 in daylight and 6 under tungsten, unchanged, because tungsten remains reasonably balanced through the green band. The same pattern holds on the slower Tri-X 320: its No. 11 yellowish-green sits at factor 4 daylight and 4 tungsten, while a No. 29 deep red swings from 16 down to 10 across the two sources. A green filter is the honest counter-case: it tells you the source has not abandoned the middle of the spectrum, only starved its ends.

Sky darkening and contrast, in zones

Tonal control is not only about sky tone; the filter alters the contrast index of the negative. The general rule from Ansel Adams’ The Negative is that red filters raise effective contrast above normal, blue filters lower it, and green filters render roughly normal contrast. For the sky specifically, against an unfiltered baseline a light yellow darkens a blue sky by about half a stop, an orange (#21/#22) by about a stop, a No. 25 (A) red by about 1.25 stops, and a very deep red (#29/#92) by up to about 1.5 stops, with the total useful range running to roughly three zones. Adams reached for the Wratten No. 29 deep red when he wanted his darkest, most dramatic skies. All of these figures assume blue is present to subtract; under tungsten, where it is not, a red filter that would drop a daylight sky two zones leaves an already-dark tungsten tone almost untouched.

Datasheet figures are midday approximations

Manufacturers treat their published factors as conditions, not constants. Ilford’s FP4 Plus datasheet notes that in late afternoon or winter, when daylight itself contains more red, green and blue filters may need slightly more exposure than the listed daylight factors, because those numbers assume average midday light. Ilford’s own ORTHO Plus tables carry the principle to its conclusion with two full columns: the 104 Alpha yellow is factor 2.5 in daylight but 1 under tungsten; the 109 Delta deep yellow is 5.5 / 3; the 304 tricolour blue runs 3 / 5; the 404 tricolour green 8 / 4.5. These are Ilford’s own filter designations, not Wratten numbers — the 304 blue is not a filter you order by that code elsewhere, it is a line in Ilford’s table. The same film records ISO 80/20° in daylight but only ISO 40/17° under tungsten, a full stop slower, because so much of the tungsten output falls in the red the orthochromatic emulsion cannot see; 135 cassettes are DX-coded ISO 80, so for tungsten work you set ISO 40 by hand or dial in a one-stop correction.

The practical escape from all of this is to meter through the filter. A through-the-lens meter shares the film’s spectral response closely enough that it sees roughly the same reduction the emulsion does, which sidesteps the factor tables entirely whenever the light is uncertain. The tables remain the reference for handheld metering and for understanding why the correction changes, but the meter behind the glass does the source-dependent arithmetic for you.