The motor innervation of cat spindles was examined in hindlimb muscles using a variety of techniques employed in light and electron microscopy. Observations were made on teased, silver preparations of 267 spindles sampled from the peroneal, flexor hallucis longus, and soleus muscles, hereafter referred to as the PER /FHL /SOL series. The
y innervation
. Trail endings are almost invariably present, and innervate both bag and chain muscle fibres. T rail fibres accounted for 64.6 to 74.8 % of the total fusimotor supply to samples of spindle poles in the PER/FHL /SOL series, the m ean num ber of fibres per pole varying from 2.7 to 5.0 in the different muscles, and the mean number of ramifications (areas of synaptic contact) per fibre being 3.7. By contrast, the p
2
innervation of a spindle pole generally consists of a single fibre supplying only one plate. In the above samples p
2
fibres accounted for 4.1 to 28.0% of the total fusimotor supply, and the mean number of fibres per pole varied from 0.3 to 1.2 in the different muscles. Ninety per cent of p
2
plates innervate bag fibres. The
α innervation
. The structure of p
1
plates as seen in both light and electron microscopy compares very closely with that of extrafusal plates. After nerve section p
1
plates degenerate at the same time as extrafusal plates, being the first of the three types of fusimotor ending to disappear. The frequency of the p x innervation is similar to that of the p 2 innervation. In the same samples of P E R /F H L /S O L spindle poles as above p x fibres accounted for 6.0 to 28.8 % of the total fusimotor supply, the mean number of fibres per pole varying from 0.25 to 2.1 in the different muscles. The majority of p
1
fibres enter a pole to terminate in one plate only. Seventy-five per cent of the plates innervate bag fibres. The three types of fusimotor ending are thus not selectively distributed to the two types of intrafusal muscle fibre. All three types of fusimotor fibre may branch within the spindle so as to innervate both bag and chain fibres. Bag fibres receive both types of plate ending as well as trail endings. Most chain fibres receive trail endings only; the rest receive either a p
1
or a p
2
plate innervation in addition, 25 % of the p
1
and 10% of the p
2
innervation being distributed to chain fibres. The significance of this nonselective innervation is interpreted as indicating that the type of contraction elicited by stimulating a fusimotor fibre depends upon the type of ending initiating it rather than upon the type of muscle fibre executing it. Reasons are given for concluding that the dynamic response is controlled via the p
1
and p
2
plates, and that the static response is controlled by the trail endings. The participation of the a fibres in mammalian fusimotor innervation, previously regarded as a vestigial feature, proved to be widespread in the muscles studied and more prevalent in fast muscles (FHL, peroneus digiti quinti) than slow (soleus). A low frequency of p
1
innervation is offset by a high frequency of p
2
(as in peroneus longus), and vice versa (as in FHL). It is unlikely that collaterals from slow a fibres innervating type B muscle fibres are wholly responsible for the high frequency of the p
1
innervation in FHL, and it is suggested that collaterals may also be derived from fast a fibres innervating type C muscle fibres. The possibility of there being some motor fibres of a conduction velocity and with an exclusively fusimotor distribution is also taken into account.
The absence of centrioles from spindle poles of rat kangaroo (PtK2) cells undergoing meiotic-like reduction division in vitro.