Ah, most important, you have to define the context! It's not enough to put the apostroph ;)

ignore [@letter @start_swash] @letter';
sub @letter' @letter by @start_swash;

for example...

try this

swsh {
ignore sub @ALL @swsh1';
sub @swsh1' by @swsh2;
} swsh;

where

@ALL = All letters and punctuation
@swsh1 = The letters you want to be replaced by swashes
@swsh2 = swash variants

Have you tried renaming your classes?
I know this suggestion might sound silly, but you never know...

One other thing, have you tried generating a font and test it in InDesign, for example? Does it work?

as you have things set up now, you need to include all your x.start letters in your @letter class. sorry i didn't make this note earlier (i missed the jpg image somehow)

Maybe the first ignore also prevents the end-swash substitution. By defining separate lookups, you also define the boundaries for each "ignore" (only valid within the lookup in which it was defined).

feature calt (

lookup caltstart (
   ignore sub @letter_start @start';
   sub @start' by @start_swash;
) caltstart;

lookup caltend (
   ignore sub @end' @letter_end;
   sub @end' by @end_swash;
) caltend;

) calt;

Mr Hunt's suggestion to include X.start glyphs in the @letter class may also applies for X.end glyph substitution.
Either you include X.end glyphs in @letter too. Or you create separate context-classes for caltstart lookup (including a-z and a.start-z.start) and for caltend lookup (including a-z and a.end-z.end); I indicated this by adding "_start" and "_end" to "@letter" in the sample above. Both may result different substitution behavior with one/two-letter-words, so you should try both ways and see which looks better.

That was it! Thanks Paul and Karsten! There was one small thing that needed changing. Instead of ( ) it needed to be changed to { } on your code.

/track/