While some caps had their outer foil clearly marked (VCap, Hovland, and so on), the RelCaps (among others) were sometimes a problem. When we bought RelCaps from Michael Percy Audio, the outer foil was already marked by MPA. Otherwise, we used an oscilloscope to locate it ourselves. To do this, we attached the hot and ground wires of the scope's probe to each lead of a given cap and placed the cap perpendicular to an active AC line, such as the power cord of the scope, itself. We observed the noise pattern on the scope, reversed the leads, and determined that the correct orientation was the one in which the scope's pattern display was the shortest in height. That is to say, when the display was the shortest, the scope's probe ground connection was considered to be connected to the outer foil of the cap.
Notice they say "... the scope's probe ground connection
was considered to be connected to the outer foil of the cap". This is their out. All they're really saying is that they had a method or process for making a decision, and stuck to it. They're not actually saying say it's the correct method, or a usable method, or even that it gives accurate /consistent / repeatable results - simply that they had a method, and the
method itself was consistently applied.
It's both a blessing and a curse. On the one hand, it's a nice easy point for a later researcher to criticise - but, on the other hand, they then have to go through the whole process of proving the method is rubbish (and face the inevitable arguments with people who claim it isn't) before getting to present their own findings. And in the meantime there'll be plenty of other lazy studies which simply state something like "Our study used the method initially proposed by Bullshyte and Crapp, which was considered suitable for determining which lead is connected to the outer foil".
You might be surprised to learn how often that particular manipulative construct is used in scientific literature...