I fomerly assumed
*Hwer- is a PIE B (in my own terminology) root corresponding to Paleo-European
*par(r)- and PAltaic
*p`árà (n) 'cross-beam, constructing place'. However, the proposal by Ernout-Meillet of linking Latin
parie:s to the root
*twerH- 'to keep, to hold, to fence' found in Balto-Slavic, assuming
*tw > Latin
p opens a new field for comparative work, as IMHO this would be a PIE A root reflected in reflected in Latin
obtu:ro: 'to block, to stop up' and Romance
*at-tu:ro: 'to stop', and which I link to PAltaic
*t`jù:ru 'to hold, to obstruct'.
In my model, Eurasiatic
*tH (reflected as PAltaic
*t`) would give PIE A
*t and PIE B
*H. Then we've got:
PAltaic
*t`jù:ru 'to hold, to obstruct'
PIE A
*twerH- 'to keep, to hold, to fence'
Latin
obtu:.ro:, Romance
*ad-tu:ro:PIE B
*Hwer- 'to cover, to close' (with various suffixes)
Romance
barra 'bar, barrier',
*berruculu- 'bolt'
On the other hand, in the lexical group represented by Altaic
*p`árà (n), which designates some type of latticework, we can include Paleo-European
*par(r)- and
*waranda:, the later with a lenition
*pH > *p\ > *w. These two would respectively represent the PIE A and PIE B types, although in this case there're no actual PIE reconstructions.