Bug in TI-83 Plus?

On my TI-83 plus emulator, I input the following in the statistical command 2-SampTInt:

x1 = 51.71
s1 = 0.79
n1 = 10
x2 = 136.14
s2 = 3.59
n2 = 10
C-Level: .95
Pooled: No

The calculator returns 9.896 degrees of freedom; it should be more like 12 degrees of freedom according to Welch’s formula (which is also implemented with T.TEST in Microsoft Excel).

I’m assuming the TI has some non-standard way of computing the number of degrees of freedom, but I haven’t the faintest idea what it could be. What’s odd is that the number of degrees of freedom appears to be computed with Welch’s formula when using the parallel command 2-SampTTest.