@ScruffR - Right - The analog sensor data I was getting off the Grove Temperature Sensor v1.2 is off by a substantial margin. More than the spec suggests is possible.
A bunch of additional searches found this Particle forum discussion. It mentions that the readings from the Particle devices will go from 0-4095 instead of the arduino’s 0-1023. That’s the extra resolution (extra bits) on the analog to digital converter. The calculation for resistance needs to take that into account. I can’t find that documented anywhere but it was the clue I needed.
I hacked at the Grove Temperature Sensor v1.2 example code and fiddled with the math. It seems to be reporting a reasonably accurate temperature now.
Thanks.