Arecibo observations of the conal triple pulsar B1918+19 at 0.327- and 1.4-GHz are used to analyse its subpulse behaviour in detail. We confirm the presence of three distinct drift modes (A,B,C) plus a disordered mode (N) and show that they follow on
e another in specific cycles. Interpreting the pulsars profile as resulting from a sightline traverse which cuts across an outer cone and tangentially grazes an inner cone, we demonstrate that the phase modulation of the inner cone is locked to the amplitude modulation of the outer cone in all the drift modes. The 9% nulls are found to be largely confined to the dominant B and N modes, and, in the N mode, create alternating bunches of nulls and emission in a quasi-periodic manner with an averaged fluctuation rate of about 12 rotation periods ($P_1$). We explore the assumption that the apparent drift is the first alias of a faster drift of subbeams equally spaced around the cones. This is shown to imply that all modes A, B and C have a common circulation time of 12 $P_1$ and differ only in the number of subbeams. This timescale is on the same order as predicted by the classic {bf E}$times${bf B} drift and also coincides with the N-mode modulation. We therefore arrive at a picture where the circulation speed remains roughly invariant while the subbeams progressively diminish in number from modes A to B to C, and are then re-established during the N mode. We suggest that aliasing combined with subbeam loss may be responsible for apparently dramatic changes in drift rates in other pulsars.
We present a single-pulse study of the four-component pulsar J1819+1305, whose ``null pulses bunch at periodic intervals of around 57 times the rotation period. The emission bursts between the null bunches exhibit characteristic modulations at two sh
orter periodicities of approximately 6.2 and 3 times the rotation period, the former found largely in the two outer components, and the latter only in the first component. Many bursts commence with bright emission in second component, exhibit positive six-period drift across the full profile width, and end with 3-period modulation in the leading component. The 57-period cycle can be modelled geometrically as a sparsely filled subbeam carousel with nulls appearing whenever our line of sight intersects a circulating empty region. This interpretation is compatible with other recent evidence for periodic, carousel-related nulling and appears to support the physics of a polar-gap emission model for ``drifting subpulses, but the subtle structure of the emission bursts defies an easy explanation.
