By "in series", do you mean linearly? Depends on context.

For Commercial & Industrial (C&I) applications like rooftops, canopies, façades, canals, floating islands, and the like, panels can be strung in series linearly sure. But because C&I arrays also tend to be arranged in polygons and more often rectangles, this allows the solar designer to string in non-linear ways.

One way to string in these contexts is to prioritize loops where the start and finish of the string are 1 module apart. A lot of RF enthusiasts don't like it when modules are strung this way because conductors arranged in a loop act as an antenna that can send and receive EM waves.

Another way to string is the snake or zigzag method where strings are patterned like this: ,,,,|'''''''|,,,,|'''''''' in rows that are 2-modules wide. The snake method can also be applied to 3-mod, 4-mod, etc. wide rows depending on what stringing method ends up being feasible.

For ground-mount applications at the community-, Distributed Generation- (DG-), and utility-scales, solar stringing is usually done linearly or in loops due to the constraints of the racking, although snake/zigzag can apply in some instances. Ground-mounts can either take the form of fixed-tilt (FT) racking, single-axis trackers (SATs), or dual-axis trackers (DATs).

With FT, the usually racking setup is 2-in-Portrait (2P) where you have 2 rows of modules abutted next to each other facing South. These are usually designed in long rows along a property line, so it's easy to make strings completely linear or half-loops where you turn the string around at the halfway point of what would otherwise be a string line. FTs can also be arranged as 3- or 4-in-Landscape (3L or 4L) which allows for more loops and snakes/zigzags.

With SATs, you're always stringing things linearly.

For DATs, you get the same benefits as C&I, 3Ls, or 4Ls because each DAT is separated from the rest of the array so the mini-array can track the sun. Lines, loops, or snakes/zigzags work, with a preference for loops and snakes/zigzags because DATs are rarely sized large enough to accommodate the entire widths of strings (sometimes 15-30 modules in length).

With railroad-based solar, my bigger concern is that since the modules have to be in a single line that CAN'T repeat in rows to the North/South in a typical array, this means all those strings will need to run a long way before being collected at an inverter to turn into AC. That long length adds to voltage drop in the circuit, which is an energy loss on the system. You can get away with this by adding microinverters for every 1 or 2 modules. Microinverters like that add a lot more complexity in terms of Operations and Maintenance (O&M) purely because there are more parts to break down. With string inverters and even central inverters, you have less O&M breakage but the flip side is that those work better with more centralized/rectangular-like arrays.

Doesn't mean it can't be done, but it's certainly more expensive than any other implementation of solar.

One of the benefits I can actually see with using microinverters over string or central inverters is that you break up the "array" of modules so that shading from the trains affects less of the "array" than if everything was collected at the DC level. You can isolate the shade-affected parts better, and promote better energy reduction.

Idk I'm interested to see what comes from this! Definitely a wild idea haha

And if the railway track is electrified

We got 100% electrification rate in Switzerland!

But usually 15kV AC with 16.67 Hz, so probably not useful for connecting the panels to be honest.