In article ,
Ragnar wrote:

"Rob" wrote in message
...

This thread raises an interesting weed-identification issue:
most gardeners want to be able to identify weeds as early as possible
in their development, which normally means BEFORE flowers have formed.
However, most systems (books and websites) rely on flowers for
identification.

One of my best wild flower books is "the Wild Flower Key" by Francis Rose
(Frederick Warne 1981) which features a useful vegetative key to identify
plants not in flower. An updated edition is still in print (try Amazon).

Back in the days before punched cards were associated with International
Business Machines, Incorporated, they were used for that purpose. You
had some special ones, pushed knitting needles through, and picked out
the cards that matched.

Some 25 years back, I was slightly involved with an attempt to revive
Rothamsted's then 25-year old diagnostic key program (Genkey). It
wasn't the fact that it had been converted to Fortran 66 and its
style showed its earlier ancestry that was the problem, but the fact
that its interface was SO very 1950s!

It would be trivial to write a fairly decent search program, given
availability of suitable data. It would also be trivial to turn the
very hierarchical classifications in various books into a suitable
form for searching by any combination of keys. And even to have
the program prompt with other keys to look for to disambiguate the
plant in hand.

But the problem is getting the data in an accessible form :-(


Regards,
Nick Maclaren.