Reciprocity Failure: Why Long Film Exposures Need Extra Time

Your meter says eight seconds, so you give it eight seconds, and the negative comes back thin. That is reciprocity failure: past roughly a second of exposure, film stops responding to time in a linear way, and the correction you need grows faster than most people expect. At thirty metered seconds some stocks want four minutes.

What reciprocity actually means

The reciprocity law says that exposure is the product of intensity and time, and that the two are interchangeable. Half the light for twice as long should give an identical result. This is the assumption every light meter is built on, and for normal shutter speeds it holds up beautifully — f/8 at 1/125 really does match f/11 at 1/60.

Reciprocity failure is where that trade stops working. On film it appears at both extremes: very long exposures, typically past one second, and extremely short ones below about 1/10,000 second. The long end is the one that affects real shooting, so that is what people mean when they say reciprocity failure without qualifying it.

Why film does this

A silver halide crystal has to absorb several photons before it forms a stable latent image speck that will develop. Those absorptions have to accumulate faster than the intermediate states decay. Under bright light the photons arrive in a rush and the speck forms easily. Under very dim light they trickle in, early absorptions relax back before the next one arrives, and a proportion of crystals never reach the threshold at all.

The result is that the film behaves as if it has a lower ISO the longer the exposure runs, and the effect compounds. It is a property of the emulsion chemistry, which is why different stocks fail at different rates and why the manufacturers publish correction tables.

The power law

The standard model is a simple power relationship, the Schwarzschild law:

t_corrected = t_metered^p

where p is the Schwarzschild exponent for that emulsion. A perfect film would have p = 1.0. Real black-and-white stocks sit between about 1.1 and 1.6. The exponent looks harmless until you notice it is an exponent: at ten metered seconds, p = 1.3 asks for 20 seconds, while p = 1.6 asks for 40.

Metered	p = 1.16	p = 1.31	p = 1.62
1 s	1 s	1 s	1 s
2 s	2.2 s	2.5 s	3.1 s
5 s	6.6 s	8 s	13 s
10 s	15 s	20 s	42 s
30 s	51 s	83 s	3 min 54 s
60 s	114 s	3 min 25 s	12 min

Note how little correction is needed below about two seconds, and how brutally it escalates past thirty. This is the practical shape of the problem: the difference between metering four seconds and metering forty is not a factor of ten in exposure time, it is closer to a factor of thirty.

Stock behaviour varies enormously

Some emulsions are engineered specifically to resist reciprocity failure — those aimed at astrophotography and technical work historically were. Others, particularly cheaper traditional-grain stocks, fail hard. Two things follow. First, never apply a generic correction table to a stock you have not checked; the spread between the best and worst common films is a factor of three at thirty seconds. Second, if you shoot long exposures regularly, choosing a well-corrected stock is a bigger win than any amount of arithmetic.

Manufacturers publish either an exponent, a correction table, or a graph in the datasheet. Where a manufacturer gives an explicit table, use it in preference to the power law — real emulsion behaviour is not a perfect power curve, and the published figures are measured rather than modelled.

Colour film has a second problem

Colour negative and transparency film carries three emulsion layers, and they do not fail at the same rate. As the exposure lengthens, the layers drift out of balance and you get a colour cast that gets stronger with time — usually toward green or cyan on transparency stock. This is called reciprocity crossover.

Correcting it needs colour compensating filters on top of the time correction, and the filter pack changes with exposure length, which is why the datasheets for colour film give you two columns. On negative film a mild crossover can be corrected in scanning or printing. On transparency it is baked in, which is one reason long-exposure work on slide film has a reputation for being unforgiving.

Digital sensors do not do this

Silicon has no latent image to decay, so a digital sensor obeys reciprocity essentially perfectly across any exposure length you will use. If your meter says eight seconds, eight seconds is correct.

Digital long exposures have their own problems, but they are different ones: thermal noise accumulating in the sensor, amp glow at the frame edges on some bodies, and hot pixels. The countermeasures are dark-frame subtraction and stacking shorter frames, not exposure correction. Anyone moving from digital to film for long-exposure work tends to get caught by this exactly once.

A practical workflow

• Meter normally and note the indicated time.
• If the indicated time is under one second, shoot it as metered.
• Past one second, apply the correction from the datasheet or the power law.
• Apply filter factors before the reciprocity correction, not after — a 10-stop ND turns a 1/60 metered reading into 17 seconds, and it is that 17 seconds that gets corrected, not the 1/60.
• Bracket long. At the four-minute end, a stop either way costs you a lot of time but very little else, and the correction models are approximations.

That fourth point catches people constantly. The order of operations matters: filter first, then reciprocity, because the reciprocity correction is a function of the actual exposure time the film sees.

Related reading

For the medium comparison behind all of this, see Film vs Digital. The filter side is covered in ND Filters and Polarizers and Fixed vs Variable ND Filters, and the exposure fundamentals in Aperture and F-Stops Explained and ISO and ASA.