SPSS linear mixed model for repeated measures at two levels

I'm struggling with a basic question for which I couldn't find an answer in the texts and papers on LMM in SPSS.

This is the basic structure of the dataset:
- outcome = grip strength
- effect of interest = injury on dominant vs non-dominant side

The source of repeated measures arises at two levels:
1. Every subject is measured at 7 different positions of the arm.
2. There are 2 attempts for each position.

I've started modeling via Analyze->Mixed Models->Linear. In the initial dialogue box, I've obviously entered Subject-ID into the box "Subjects".

The main question is: should I also enter "position of the arm" in the Subject box?

Hopefully someone knows the answer to this basic question.

For completeness, this is what I filled in at the other fields:

Initial dialogue box:
Repeated measures: Attempt
Covariance structure: Compound Symmetry

Main dialogue box:
Dependent variable: grips strength
Factor: injury on the dominant vs non-dominant side
Covariate: position of the arm (this is actually a scale type variable, but with 7 fixed positions!)

Fixed button:
Fixed effect:
- injury on the dominant vs non-dominant side
- position of the arm

Random button:
- Model: position of the arm
- Subject grouping: Subject ID
Furthermore, I included the intercept and chose autoregressive as a covariance structure.