I think the short value incoming into the binding is 1/1000th of an inch of mercury, but it’s being treated as an inch of mercury. Multiplying an inch of mercury by 33.8638 produces hectopascals, so multiplying a thousandth of an inch of mercury by 0.0338638 also produces hectopascals.
@richard_altherr,
Could you replace your Davis binding JAR with this one and confirm or refute the idea that the unit conversion is working correctly now?