Workspace planning
To conclude this section, it is worth remarking that care taken in structuring workspaces will often pay dividends if the analysis results need to be returned to later. An Explorer tree represents a single workspace. It can contain several starting data matrices, the properties (factors etc) of each being accessible to the others if they are based on an overlapping set of identical sample labels. (As seen previously, factors or indicators associated with a particular data matrix are auto¬matically available to other sheets on the same branch, and on different branches by using the Factors> Import button). Such is the power and reach of PRIMER to quickly generate many plot and results windows that the user will probably find it a constant battle to keep workspaces down to a manageable size, both from the viewpoint of ease of navigation around them and of the size of a workspace file that needs to be transmitted to others (we have repeatedly seen the convenience of saving all the information in the workspace in a single file with a File>Save Workspace As step).
Firstly, it makes obvious sense to keep analyses on different studies in different workspaces but, also, analyses on different components of a single study may sometimes be better carried out in separate workspaces. The key criterion for using a single workspace is whether two data sets are to be combined or input to an analysis step together (e.g. species and matching environmental arrays), even if this is just as component parts of a single multi-plot. If not, they are probably best held and saved as separate workspaces. Multiple launches of the PRIMER desktop are straightforward, each with a different workspace (if the same workspace is opened twice – which is perfectly possible though should usually be avoided because of the likelihood of confusion when saving! – the second will be a copy of the saved version of the first workspace). Parallel desktops will not interfere with each other, and are never linked in any way. The only means of transfer between them is by saving individual sheets (e.g. data as *.pri) and then opening (a copy) of that file into the other workspace.
Secondly, ‘housekeeping’ within a workspace is important for intelligibility: key sheets of data, resemblances, results or plots – anything that needs to be selected as the active sheet for a further analysis, or a result or plot window that has been copied and saved to an outside presentation or manuscript – should be renamed in a meaningful way (all names in the Explorer tree need to differ, though PRIMER will ensure this by adding (2), (3), .. to the end of any name you supply which is already used in the workspace). Also, it is usually advisable to delete clutter, e.g. analyses you now realise were flawed or sub-optimal, using File>Delete … or Delete Item on the right-click menu when the window to be deleted is highlighted (branch entries below that will be deleted too). Use File>Undo Workspace if you make a mistake and want the excised portion back again! If you decide the sub-optimal analysis needs to stay in, as a reminder, then roll-up that particular branch (with 
) and attach a note to the window above the icon, with right-click Add Note. In fact, it is desirable to make good use of this annotation feature more generally, to aid navigation.
Finally, some studies are sufficiently extensive, with data accreting over time, that it is advisable to resave the workspace with a modified name from time to time, so that at least an earlier version can be returned to should a disaster happen to the current workspace! Or it might be efficient to save the current data matrix (or matrices) in PRIMER binary format, *.pri, thus retaining all existing factors and indicators, and re-open this in a clear workspace for the next phase of the analysis.