> I guess you are saying that it is up to the actual user of the components to
> measure their characteristics. For me, there is only one problem with this,
> I am at this point only an electronics hobbist. I am studying to become an
> electrical engineer. I don't have the money to buy the test equipment needed
> to measure all the characteristics I need to know.

I think what he's saying is that the design should be such that those
parameters DON'T MATTER.  While some transistors may be engineered so
that they behave very precisely according to a detailed specification,
it's much cheaper to make them "sloppier"; while the sloppy transist-
or will require some extra circuitry to cancel out any slop, the extra
circuitry will cost less than making the transistor less sloppy.

As a simple example, suppose you're trying to design a circuit which
will convert a positive voltage to a negative current.  There are two
ways of doing this with a single transistor:

Method 1:
           /----- out
         |/
In-/\/\/-K
         |\
           \|--gnd

Method 2:
          /----- out
        |/
In------K
        |\
          \|--/\/\/--gnd

In both circuits, the current sunk by "out" will be proportional to the
voltage at "in" minus Vbe.  In the former circuit, however, the current
is EXTREMELY sentitive to the parameters of the transistor, while in the
latter case it is MUCH LESS sensitive.A  Further, in the former case, the
accuracy of the design requires that the transistor's beta be set to a
precise value, while the latter merely requires that it be as high as
possible.

If the accuracy obtained by "Method 2" above isn't adequate, then it may
be improved further by using op-amp feedback circuits.  While op amps used
to be expensive (esp. before single-chip monolithics came on the scene)
they are now very cheap; using a circuit with op amps and feedback, it's
possible to design an amplifier or control circuit whose accuracy depends
[for all intents and purposes] solely on the accuracy of the resistors and
caps that make up the design--not on the specific properties of the trans-
istors.

To bring things back to the world of PICs, consider the specifications of
the PIC's port pins; earlier PIC books had graphs detailing how the PIC's
ports would sag when confronted with large current loads.  If one took these
graphs at face value, it would be quite safe to hook up two LED's in series
between two port pins with no resistor; since the VDD rail would drop by
about a volt under a 20mA load and ground would rise by about half a volt,
about 3.5 volts would remain for the LED's.  Any design that was built that
way would have problems, though, if Microchip were to "improve" their PICs
by beefing up the I/O ports.  By contrast, using two LED's each driven by
its own port pin and with its own current-limitting resistor would be a
much more "stable" approach.  While the LED's might get a smidge more
current on newer PICs which have a smaller voltage drop, the difference would
not be very big.  And even if a future model PIC came out which had NO volt-
age drop on its output high/low, the circuit would still be safe.