The caster, when he casts the spell shatter in its first mode, designates a point of origin for the spell within the spell's range. Within the spell's area, each object that is both unattended (i.e. not carried, held, possessed wielded, or worn by a creature) and appropriate (i.e. nonmagical and made from "crystal, glass, ceramic, or porcelain") is destroyed, and each object that is both appropriate and attended makes a Will saving throw using the saving throw modifier of the creature attending the object; failure means that object's destroyed. Everything else is unaffected.
This does, however, make the shatter spell's saving throw entry for its first mode a little wonky.
The (object) portion of shatter's first saving throw entry is unusual
When the shatter spell's used as an area spell, it probably would've been enough for spell's designer to've listed the saving throw as Will negates then let the spell description do the heavy lifting of detailing what that means rather than asking the reader to pointlessly look up the details of the rarely-used (object) tag, which, in this case, brings little to table. That is, when used as an area spell the spell shatter isn't cast on the objects but on the area itself, so the saving throw here entry of Will negates (object) is pretty much the same as Will negates.
Except that a saving throw entry of Will negates implies that the creature makes just one lone Will saving throw against the entire effect of the spell shatter, with success leaving unharmed all the creature's appropriate stuff and failure rendering broken all of the creature's appropriate stuff. To clarify that isn't the case and that the creature makes saving throws for its attended appropriate mundane objects individually, the author would've then had to've said as much in the spell's description. Thus, here, the (object) tag becomes more reminder ("Make a separate Will saving throw for each appropriate object in your possession!") than rules ("Do the impossible and make several objects—rather than the area—the subject of an area spell!").
I agree that it would've been great had Pathfinder seized the opportunity to revise some of the game's classic elements—like the spell shatter—so that they were absolutely (ahem) crystal clear, but the game contented itself with functional—if imperfect—legacy spells dating back to the turn of the century that had been working fine as printed for over a decade,—in this case, not fixing what isn't outright broken.