I've made a small hexagonal array of 7 transducers spaced 2Lambda apart (1 in the middle and 6 around) and it narrows the beam width from 70 degrees to about 30 degrees. Curiously there is a dip on the bore-sight.
Did you make sure all the transducers are connected in the same phase?
Are the fronts of all the transducers flush with each other, in a plane perpendicular to the beam?
How far away from the transducers did you measure the beam width? If you measured it close to the transducers, you may need to set the center transducer back slightly, to equalize the distances between the transducers and the center of the beam at the measurement point, in order to avoid a dip in the centre.
It's many years since I studied physics, so I'm not sure that there isn't some mathematical reason why you should expect a dip in the center; however it doesn't sound right to me.