traverse closure question
As others mentioned, adjustment is for random error not systematic error or blunder. Also, not enough to close back on the hub, you have to set on it again and turn the final angle. Then balance angles to see if angular problem. You can only analyze that which is part of the loop, sideshots simply get calced after removing systematic error (such as prism constant problem) blunders (such as kicked the hub before backsighting) and adjusting random error of the loop. Sideshot coordinates will then be based on the corrected coordinates of the loop points, but there's no check on whether you got correct angle and distance to individual sideshots unless you already had a coordinate for them to compare. The points you re-sected from are not part of the loop from what you describe, so wouldn't have anything to do with the adjustment. If there's a problem with the points you re-sected from the whole thing will just be off by whatever even if you get the loop to close perfectly, but the traverse analysis will not tell you one way or another.
Close the traverse by turning final angle. Then start checking angle and distance on each traverse hub till you find mistake (not error). Small site like that shouldn't be too difficult. Large site least squares can help find where to start checking.