The original and revised balance-point models of flagellar length control and possible mechanisms of IFT train size modulation. (A) The original balance-point model predicted that flagella contain a fixed number of IFT trains. Thus, as flagella regenerate, the length-dependent assembly rate would be driven by the decreasing frequency of train arrival at the flagellar tip. This model was refuted by the DIC observation that IFT frequency is constant (Dentler, 2005). In the revised model, the length-dependent assembly rate is mediated by IFT train size that scales inversely with flagellar length. Although the total amount of kinesin-2 and IFT protein in a flagellum is length independent, these proteins are redistributed into a greater number of smaller trains as the flagellum regenerates, reducing the rate of flagellar assembly. (B) Two possible models of train size control. In a closed system, at least one essential component of the IFT machinery does not exchange freely with the cytoplasmic pool. As the flagellum lengthens, lower concentrations of this key protein arrive at the flagellar base via retrograde transport, which results in the production of smaller anterograde IFT trains. In an open system, there is always a high availability of IFT material at the flagellar base, as proteins freely exchange with the large cytoplasmic pool. Thus, an additional length-sensor mechanism is required to modulate the size of trains that enter the flagellum. Blue arrows indicate high IFT protein concentration, whereas red arrows indicate low concentration.