CCL: One unpaired electron - Zero integrated spin density?

[Stephen.Garrison*|*srnl.doe.gov]

Could someone explain to me how a system
 can have one unpaired electron, but nearly zero integrated spin density?
 
 
I am using CASTEP (LDA functional) and
 am working on a periodic metallic system with 189 electrons. 
 
 
I start the calculation with an estimate
 of 5 unpaired electrons. The SCF convergence cycle for the initial geometry
 finds a solution with the following spin densities:
 
 
 2*Integrated Spin Density  
 =   -0.533909E-01
 
 2*Integrated |Spin Density| =
    0.972315    
 
 
The final, optimized geometry has the
 following spin densities:
 
 
 2*Integrated Spin Density  
 =   -0.337735E-03
 
 2*Integrated |Spin Density| =
    0.612583E-02
 
 
This does not make much physical sense
 to me. How can a system have an unpaired electron but zero integrated absolute
 value(spin density)? What am I missing?
 
 
Thanks for you help and clarification.
 
 
Stephen
 

 --
 Stephen L. Garrison, Ph.D.
 Senior Engineer
 Engineering Modeling and Simulation
 Savannah River National Laboratory
 Savannah River Site
 773-42A, Room 181
 Aiken, SC 29808
 803-725-2404
 Stephen.Garrison]|[SRNL.doe.gov