After a superficial and really fast search in the web, effectively elm seems to be a poor studied species under dendroarchaeological point of view, or at least free-access data seems to be hard to obtain. I found just one sample of elm from Italy (http://dendro.cornell.edu/reports/report2000.pdf
): sadly it’s a protohistoric sample and furthermore it’s not dated because of some problems during growing. Anyway maybe you can find some curves googling in different languages and searching in real libraries.
If you don’t have a reference curve of the species you are interesting in, here goes, just as suggestion, a personal, absolutely inelegant, “protocol”:
1) Search for at least one dated curve of elm from northern Italy, Alps, etc… It have not to be necessarily in digital format: you can convert published diagrams by using the coorecorder facility. Look also for ecological-related papers: often they have short curves, too short for dating purposes but useful by the point of view of this personal method. If you don’t find any, sample yourself a little curve.
2) Search in all databases (two more: http://dendrodb.cerege.fr/
) and in published papers all the curves near the zone from which is your little elm dated curve.
3) Play with the data for revealing better correspondence between elm and other species in the same area.
4) Do a “massive attack” comparing all curves you have collected with your undated curve (you can find examples of the process in Lars-Ake’s site, under "Miscellaneous" section), evaluating the results by aid of observations you made after point 3. The dated reference curves, the little dated elm curve and the undated elm sample have to be from the same area.
5) If you are lucky enough, you can obtain some useful deduction and maybe some result too.
6) Since this method is not rigorous and can bring EASILY to incorrect datings, do further and endless verifications. If you are starting now from zero, you can spend years until you can date reasonably the sample.
Anyway, you have to be sure your undated sample is elm, and it aids if you know approximately when the sculpture you are trying to date was made. Probably you already took in account this, but I would pay attention to the context too: may it be an imported sculpture? From where?
Document yourself about dendrochronological characteristics of elm, even from other regions, especially looking for interspecific correlations values (sorry, I never touched elms data). Finally, you probably already know that possibilities of correct dating are proportional (among others parameters) to the number of samples forming your undated curve.
Hope this is sufficient to start. I’m sorry if this reply looks too doctrinaire (and surely I’m not the adequate one...), but you told you was starting with dendrochronology. It’s just a suggestion and the process described could be refined. You can have an overview of Italian dendrochronology status (and, by the way, you can convert to digital data some published long chronology) reading papers by Franco Biondi (search the web for pdf of him), although they are from early '90 and I think elm is not quoted.