Chaps, it looks like I was too brief when I first asked the question.
The product is a PC card, the brains of which is a multifunction chip.... Two pins of this chip are for connection to an X-TAL and two tiny caps ( pF's ) to ground. The initial value for the caps was 47 pF, and the Radiated emissions from the card where about 6 dB above the Class B limit. I disabled all the circuits on the board until just the oscillator part of the chip was running, the strong spectral line emissions were about the same, still 6 dB over. Killing the oscillator stopped the emissions ( at least something happened I expected....). Thinking I may be able to sway the oscillator frequency enough to detect by increasing the value of the two caps ( actually I ended up adding 100 pF to make a total of 150 pF ), the radiated emissions dropped by about 20 dB up to about the 11th Harmonic: after this they converged just before going into the noise level of my set-up. The layout is not bad at all. It has short traces ( 1/4" max ), a small ground area ( about 1/2" * 3/4" ) surrounding the X-TAL and the two caps. The "Brain" chip has good decoupling, and the board uses dedicated ground and power layers. I've worked on many X-TAL designs ( from an EMC point of view ), but I've never seen this type of effect. Hope this is a clearer picture, Best regards, Derek.
