A smaller diameter tubing unit has less cross sectional area of gas and so when ionized it presents higher resistance than a larger diameter tube would. (this is also why you can run more 15mm tubing on a given transformer than you can 10mm) This higher resistance, combined with the low pressure at that stage of bombarding created a condition where there was some lower enough resistance through the tubulation to allow some leakage back that direction. Since what you had was not a full strike back through the tubulation, I'd almost call it a "glow back" rather than a true high current flashback. Nonetheless it is undesirable.
Contributing factors to this happening not only include the tube and pressure but can also include impurities that may be lurking within parts of the system such as on your O-rings. We usually only change them or mess with them when there is a leak, but other more subtle problems can arise before an obvious leak and this could be an example of one. Another factor is the striking voltage of your bombarder and its ability to push current through the tubing unit. Lastly, and often not considered-- the pressure within a system and its connected tubing unit is not always uniform throughout. This is especially true of smaller diameter tubulation and units where more time is needed due to the molecular flow characteristics at low pressures. Did you just previously open the main to lower the pressure, then close it and strike the arc?...if so, it is entirely possible that the section of tubulation and manifold itself were at a slightly lower pressure than the run of tubing being processed and this could result in enough of a resistance difference to lead to some current flow in those other parts of the system. Most likely you are seeing the effect of a combination of all these things.
There are several things that can be done to mitigate this, but on the transformer front--depending upon what type of unit you have (midpoint grounded or full floating secondary) we used to pump two units at once, jumpering at the close end and putting the hot leads on the far end. This worked because it meant that the available potential difference with respect to ground at the close end was reduced by the voltage drop of the unit being pumped. The trouble is, in this day and age, we rarely pump two at a time anymore. But, if your transformer has a full floating winding, you can connect a hot lead at the far end, and ground the lead at the close end, and ground the close electrode, and ground the manifold/pumps/etc all together. This makes the transformer function like an end grounded unit. (similar to the "single bushing" pole transformers) and the result is that EVERYTHING at the manifold end is all at ground potential and so no flashback can occur. BUT. This will NOT work unless you have a full floating secondary!
Give me a call if you'd like to discuss in more detail or bounce some ideas around.