Passage A:

A photograph differs from a drawing in one crucial respect: light reflected from what stood before the lens helped produce the image. That causal connection gives photographs an evidentiary force that interpretation cannot simply dissolve. Framing excludes what lay beyond the border; exposure changes what can be seen; a caption can misidentify a place. Still, if an unaltered photograph shows a banner above a crowd, then some configuration capable of producing that image existed before the camera. The image may not tell us why the banner was raised, but it constrains accounts that deny its presence.

Skepticism about photographic evidence often targets the fantasy that pictures explain themselves. No serious evidentiary use requires that fantasy. Historians compare images with testimony, determine dates from material features, and examine sequences rather than isolated frames. Such practices do not manufacture the photograph's relevance; they specify what its physical trace can support. A fingerprint also requires interpretation, yet its need for interpretation does not make it equivalent to a detective's sketch. Photographs are fallible witnesses, but their fallibility is bounded by an encounter with the world.

This boundary matters most when a claim is negative. A frame cannot establish that no banner existed outside its border, but it can refute the claim that none appeared within the recorded view. Interpretation determines the claim's scale; the trace then answers back. Evidentiary humility should narrow conclusions, not erase the difference between constraint and invention.

Passage B:

Calling a photograph a witness obscures the labor that makes it evidence. Cameras register light, but archives register significance. Someone decides which exposures to preserve, how to caption them, what event name will retrieve them, and which sequence will be treated as representative. These decisions do not merely explain a ready-made fact. They establish the object to which later claims refer. A photograph filed under “public celebration” enters history differently from the same image filed under “state-organized demonstration.”

Survival itself is selective. An official archive may preserve the photographer hired by the state while losing images made by participants, so abundance can misleadingly resemble representativeness. Before interpreting what a collection shows, historians must ask what routes allowed some views to endure.

This does not mean photographs can be made to prove anything. Material analysis can expose alteration; provenance can defeat a false date; details inside a frame can resist a caption. But resistance is not self-sufficient testimony. Even deciding that a detail matters depends on a question and a comparison assembled outside the image. The evidentiary unit is therefore not the photograph alone but a chain linking image, custody, classification, and inquiry.

The language of causal trace remains useful if it marks one constraint within that chain. It becomes misleading when treated as a complete theory of evidence. Historians do not encounter the past at the instant of exposure; they encounter selected objects whose routes to the present are uneven. To evaluate a photograph is to examine both what reached the lens and what subsequently made that exposure visible, describable, and worth consulting.