The requirement for capacitors on crystals when used with any microprocessor
varies with the circuit inside the chip. Some chips require capacitors to
oscillate, others dont. Some crystals require capacitors to oscillate, others
dont. The differences in the gain of the inverter chip inside the micro, and the
effective impedance of the crystal all combine to make life a bit of a black art
in determining just what caps are required.

having said that, most chip manufacturers have got themselves organised to
design the oscillator portion of the chip so it is pretty laid back about the
crystal characteristics, and most crystal manufacturers now make crystals which
are well enough defined in their characteristics to always oscillate with any
micro. If you find a crystal that does not work, then it is most likely at an
extreme of the characteristic of either the micro or the crystal. It can
sometimes be made to oscillate by fitting one or two capacitors to shift the
oscillation frequency enough to change the effective impedance of the crystal,
thereby getting enough phase shift around the loop.

It is possible to change the frequency of oscillation using capacitors, but this
would really only be necessary if you were wanting to use the micro for a clock,
as in the type of clock displaying the time. If you were wishing to time stamp
data collection functions I suspect you would probably find it more cost
effective to have a separate dedicated clock chip you could read, or have a
radio locked receiver listening to one of the 60khz time stations. Does anyone
know where I can get info on the UK Rugby 60khz time transmissions used by the
so called radio clocks?

To change the frequency of a series tuned crystal, the normal method is to have
both a capacitor and an inductor, which form a series resonant circuit at the
frequency of interest. By varying either the capacitor or the inductor, the
crystal sees a change in the impedance in series with it. If the series resonant
circuit is tuned so that at the nominal series frequency of the crystal, the
circuit looks inductive, then the crystal will operate at a frequency where the
crystal has a matching capacitive impedance. If the resonant circuit is tuned to
look capacitive, then the crystal will operate at a frequency where it will look
inductive.

There are various problems to using a crystal as a series tuned device, so in
the sort of circuit used for micros this is not usually done. Unless the crystal
is very carefully made, it is highly likely to try and run at an odd multiple of
its fundamental frequency, and so this almost always requires a tuned circuit
somewhere around the oscillator at the desired frequency. This is generally done
as a series resonant circuit in series with the crystal, which has the advantage
of frequency trimming as mentioned above. Some notable oscillator units that did
this sort of thing in the past were the original oscillator block for the 8080
micro, and I think the one for the 8086 also.

In short I just about guarantee that a one chip micro, or any chip with an
onboard oscillator is using a parallel resonant crystal.