The way I'd do it would be to have a big 26-light display, each corresponding to one letter. Then you'd light them up one at a time and have the receiving clerk write down the letters one by one. Efficient? Not very, but I can see how someone without a lot of information theory knowledge (as almost all early inventors would be) would land on that idea first.

There were pre-Morse / Wheatstone telegraph experiments using 26 wires. One notable early system by Salva Campillo used 26+ tubes of electrolyte, one for each letter. Closing the circuit would cause bubbles to form in the tube at the other end.

Even wackier systems from before this time used static electricity ("current" electricity was harder to generate in the early days). One proposal was to have a different servant hold a wire for each letter of the alphabet. At the other end a high-voltage Leyden jar would be connected to wires in turn, giving the servant a shock and prompting them to speak their assigned letter. Unfortunately I can't find the source for this one right now (I think it was in the book The Story of Telecommunications).

Ditto; great article.