As old mart says, if you are going to make one do not use dovetails.
A bad design concept at the best of times in this application and very demanding on alignment relative to both bed and cross-slide.
For perfect operation the dovetail way has to be parallel and perpendicular as appropriate to both lathe bed and cross-slide. On an affordable, lighter lathe, there is no guarantee the alignments are sufficiently good for the attachment to work properly unless the gibs are loose. Especially on an older machine with some wear or one not originally made to be capable of taking a taper turning attachment.
Systems using a square rail with a saddle riding on it or open topped trough with a shoe inside are free to self adjust vertically if mounted with an up hill or down hill error in relation to the bed saddel guides. Similarly the fixing pillar can be modified to accommodate any errors of cross slide alignment.
Ideally there won't be any errors but getting to within 10 thou or a degree or so its much much easier than getting the dead nuts right alignment needed to run a gibbed dovetail correctly set for free, shakeless movement.
When setting up to turn a taper its far easier to work in slope, thous per inch or whatever, than angular degrees. DRO systems and a dial gauge make this really easy. A couple of bed stops and a decently long length bar to define the travel works well enough. For imperial folks a 6" bar makes calculations easy.
Morse tapers are around 5/8" per foot, 50 thou per inch so getting real close with a 6" test length isn't hard. About 1.49° not so much. The old guys used inches per foot for a reason.
If you have nothing else a pair of toolmakers clamps gripping the edge of the bed will do for stops. But be gentle.
If you don't have a suitable dial guage fit a "bumper" in the tool post and use the cross slide dial to measure the offset at each end of the bar relative to a parallel test piece held true to the bed. Ideally between centres but holding in the chuck and turning in situ with the free end supported by a tailstock centre does well enough. Bumper needs a nicely curved end so t always contacts in the middle. A slim dowel pin driven vertically through a square bar with enough filed out of the middle so only the pin touches the test bar will do.
Clive