Certain aspects of methodology were left out of our paper owing to space constraints. Bharadwaj cites Ferrier & Thompson (Reference Ferrier and Thompson2002) when questioning the exclusion criteria used in our study. Both are co-authors of our paper, which is a result of collaborative research between the Department of Psychiatry in New Delhi and Newcastle since 1998. Whenever possible, similar tests and criteria for euthymia have been used in both centres with occasional variations to respect cultural differences. Use of spouses and siblings as members of the control group was acceptable, as it brought together people of broadly similar backgrounds. Although this might have resulted in the inclusion of a limited number of controls who were at risk of developing bipolar disorder, it minimised differences between people with bipolar disorder and controls without greatly confounding our results.

For verification of euthymia participants were seen at least twice, separated by a minimum of 1 month, before they were tested. Clinical judgement of euthymia was reinforced by a Hamilton Rating Scale for Depression score <8 and a score <20 on Bech's modification of Beigel's Manic State Rating Scale on both occasions. The stability of mood during the intervening period was assessed clinically on a weekly basis. We were not aware of any Hindi version of the Structured Clinical Interview for DSM–III – patient version. The exclusion of other psychiatric morbidity was based on clinical interviews by two highly experienced psychiatrists, complemented by careful mapping of life charts using the techniques of Post et al (Reference Post, Roy-Byrne and Uhde1998).

Soft neurological signs were assessed with the widely used modified Kolakowska battery. We are unsure whether the use of other batteries, such as the Cambridge Neurological Inventory, would substantially alter our findings. Involving a second rater would perhaps increase reliability but would extend the assessment time unreasonably.

Not surprisingly, soft signs were found in the control group, but only at about one-quarter of the severity seen in people with bipolar disorder. The maximum score on the modified Kolakowska battery was 45. The maximum score for controls was 9 whereas the mean for patients was 13.9. Control scores mainly comprised minimum scores on a few of the 15 items. In a subsequent article (further details available on request) we report high levels of soft signs in the youngest patients with bipolar disorder. There is no evidence that soft signs progress with age in bipolar disorder, whereas in controls there is significant (P<0.01) progression with age.

List A7 of the Rey Auditory Verbal Learning Test measures retention after 20 min. We have further analysed these data and found no difference between the groups.

We agree that ‘duration of illness’; actually describes ‘duration of illness episodes’. The actual mean duration of illness was 9.1 years (s.d.=6.0). Data concerning marital status and occupation were collected but were omitted for brevity. We did not wish to control for handedness or birth injuries as potential confounders as we regarded these differences to be part of the spectrum of people with bipolar disorder. We did not include those who had recently received electroconvulsive therapy (≥6 months). Finally, we would agree that there is a need for meta-analyses and have recently published such a study (Reference Robinson, Thompson and GallagherRobinson et al, 2006).