A solved problem from Griffiths creats problem.It requires some discussion...It asks to find vector potential of an infinite solenoid with n turns per unit length radius R and current I. Griffiths correlates integral form of Ampere's law to [closed integral of A.dl=surface integarl B.da].I can find A from this, with B=(mu)nI,uniform axial field and A circumferential. But Griffiths says something here which I cannot understand.Quoting him,"The present problem (with a uniform magnetic field (mu)nI inside the solenoid and no field outside) is analogous to the Ampere's law problem of a fat wire carrying a uniformly distributed current.The vector potential is 'circumferential' (it mimics the magnetic field of the wire)." As it appears, Griffiths is talking about the case where a fat wire is carrying a line current.In this case, magnetiv field is circumferential...But what is its justification?