The character they all end on

Look at the end of each of the three addresses printed above. They finish on the same symbol, and the first tile confirms it across the set.

This one is not a coincidence and it is not a choice. The material encoded into an address ends with a version marker, and because every address of this generation carries the same version marker, the encoding of that final piece comes out the same every time. Any address of this kind, from any service anywhere, ends on that symbol.

It gives you a second free check to sit alongside the length check. If a string ends on anything else, it is not an address of this generation, and you have found that out by looking at one character.

Two free checks in a couple of seconds is genuinely worth having. Between them, length and final symbol will catch a truncated paste, a string that lost its tail at a line wrap, and anything that is not really an address at all.

It cannot distinguish anything. Every valid address passes, including every address that is not the one you want. A check that everything passes filters out only the malformed.

It is worth being blunt about this because the ending has the same seductive quality as the opening. Both look like a signature. Neither is. The opening is chosen and can be copied. The ending is fixed and is shared by the entire format. The only part of the string that identifies anything is the long stretch between them.

This panel is the reason the agreement count on the divergence panel comes out higher than the shared opening alone. The final position is guaranteed to agree, so it adds to the count for a structural reason rather than a random one.

That distinction is worth keeping. Some of the sameness in the set is built into the format, some was chosen by whoever generated the keys, and some is coincidence. Only the third kind is meaningless, but none of the three kinds is evidence about the address being the right one.

• Count the characters. Anything other than 56 is wrong.
• Look at the final symbol. Anything else is wrong.
• Look for characters outside the allowed alphabet. Any of them means wrong.
• If all three pass, you have established that the string is shaped like an address. You have established nothing about which address it is.

That last line is the point of the whole strings section. Cheap checks catch damage. Only a position by position comparison against a source you already trust catches substitution, and no property of the string alone will ever do it for you.

There is one more reason the final character is worth a glance, and it has nothing to do with identity. It is the part most likely to go missing.

Long strings get truncated. A messaging app cuts a link at a length limit, a chat window wraps and somebody copies only the first line, a note taking app trims trailing characters, a paste picks up everything except the last few symbols. In every one of those cases the front looks perfect and the tail is gone.

Checking the ending catches all of it at once. It is the single cheapest guard against the most common way an address arrives broken, and it costs one glance. That is a rare ratio and worth using, as long as you remember it guards against damage and not against being handed the wrong thing.

Do older addresses end the same way?

No. An older and much shorter generation of address existed and did not carry this marker. None of the strings on this board is one of those.

If the ending is fixed, why print it at all?

Because it is part of the address and removing it would produce something that does not work. It is fixed, not optional.