You picture color channels as three separate buckets: blue light goes in the blue bucket, green in the green, red in the red. Film's actual sensitivity curves don't work that way. Plot the response of a single emulsion layer against wavelength and you get a broad, lopsided bump, not a narrow spike — and the three bumps for a color stock overlap each other across most of the visible range. That overlap isn't sloppy engineering. It's what the chemistry allows.
What the Curve Is Actually Plotting
A spectral sensitivity curve is a probability map. At each wavelength, it states the likelihood that a photon arriving at that energy will trigger a reaction in a silver halide crystal — shift an electron, start the chain that builds a latent image. High sensitivity at 450nm means a blue photon has a good chance of registering. Near-zero sensitivity at 650nm means a red photon mostly passes through unregistered. The curve says nothing yet about how the film will look; it describes only which photons get counted before development ever starts.
Stack three of these curves for a color negative and they don't sit in separate lanes. They share the same wavelength axis, and they overlap substantially — there's no band in the visible spectrum where only one layer responds.
Why the Curves Come Out Broad
A narrow, spike-shaped sensitivity would make color separation trivial: tune the blue layer to 440–460nm only, and it would never see anything else. Sensitizing dyes don't allow that. To extend a silver halide crystal's native blue sensitivity into green or red, manufacturers attach organic sensitizing dyes to the crystal surface; the dye absorbs a photon and passes the energy to the crystal lattice. The absorption spectrum of an organic dye molecule is set by its electronic structure, and those absorption bands are inherently broad — a dye that would isolate a ten-nanometer sliver of the spectrum and nothing else isn't something the chemistry offers.
The result: the blue layer answers mostly to blue but also picks up some green. The green layer answers mostly to green but leaks into both neighbors. The red layer answers mostly to red but catches some green too. Every photon that lands on the film casts a partial vote in more than one layer.
Separation Happens After Capture, Not During It
If the layers overlap this much at the moment of exposure, color still comes out distinct — but the separating work happens in the filter stack and the development chemistry, not in the sensitivity curves themselves.
Every silver halide emulsion is intrinsically blue-sensitive, which is why the blue layer sits on top of the stack and a yellow filter layer sits directly beneath it: the filter absorbs the blue light that already exposed the top layer, so it never reaches the green and red layers below to add a blue contribution they weren't supposed to register. That filter is a physical barrier baked into the film's structure, not a correction applied afterward.
Development is where the overlap gets resolved into color. The exposed silver halide reduces to metallic silver; bleaching strips that silver back out; the dye couplers built into each layer form dye during this process, in proportion to how much silver formed there. The blue-recording layer yields a yellow dye, the green-recording layer a magenta dye, the red-recording layer a cyan dye. What you see in the final image is light being subtracted through this stack of dye layers — a chemical result, not a readout of three independent channels.
Not the Same Thing as Color Balance
This overlap is a separate property from a stock's color balance. Color balance describes which illuminant a stock is designed to render as neutral — tungsten versus daylight — and it's a property of the whole system: the sensitizing dyes, the filters, and the dye couplers working together for a specific light source. The sensitivity curve is the raw physical response before any of that tuning is factored in. Change the filter layer or the dye coupler chemistry and you can shift a stock's color balance while its underlying spectral sensitivity curves stay close to identical.
It's also why grading film is a coupled problem rather than three separate ones. Because the curves overlap at capture, you can't isolate a change to "the blue channel" the way you can with three cleanly independent channels — a change in exposure shifts all three layers' silver formation together, in proportions the overlap already set. For how this connects to gamut and digital color spaces, see our piece on color gamut vs. spectral sensitivity. You can see the layer math in motion in the Cineon studio.
The three colors in a scan were never three independent measurements. They're the resolved output of three overlapping probabilities, sorted out by a filter and a set of dye couplers built for exactly this problem.