1) Yes
2) the first of your two suggestions
3) No. What you have suggested is the joint probability that both Martin and Norman are late. To get the conditional probability, you would need to divide this by the probability that Norman is late, i.e. by `P(Norman Late=T)` which would be `P(Train Strike=T)*P(Norman Late=T|Train Strike=T)+P(Train Strike=F)*P(Norman Late=T|Train Strike=F)`