A carrier effect is obtained typically when a hapten‐protein conjugate is injected into an animal which has previously been primed with the same hapten conjugated to another carrier protein. Under these circumstances the anti‐hapten secondary response is usually less than that which would have been obtained had the animal been injected with a conjugate prepared with the same carrier as that originally used for priming. Attempts have been made to account for the phenomenon in terms of the local environment hypothesis, which assumes that the receptor on immunologically competent cells recognises the hapten jointly with the area on the complete antigen which surrounds it. Alternatively the phenomenon can be accounted for by the hypothesis of cooperation, which assumes that the antigen is recognised by two receptors, one directed to the hapten and the other to a determinant on the carrier protein.

Methods are described which enable carrier effects to be studied quantitatively in mice. They involve a cell transfer system in which cell suspensions prepared from large numbers of donors are distributed among irradiated syngeneic recipients. In these recipients the transferred cells can be made to perform a secondary response by appropriate antigenic stimulation. The response is monitored by binding tests in which the capacity of serum to bind highly radioactive haptens or proteins is measured. The haptens employed in this system are NIP (4‐hydroxy‐5‐iodo‐3‐nitro‐phenacetyl‐) and DNP (2,4‐dinitrophenyl‐) and the proteins comprise chicken γ‐globulin, bovine serum albumin, human serum albumin, ovalbumin, bovine γ‐globulin, keyhole limpet hemocyanin and mouse γ‐globulin.

A carrier effect was regularly obtained when the proteins were tested against one another as carriers, for priming and for the secondary response. The effect could best be measured by comparing the relative potencies in the secondary response of the homologous conjugate (i. e. one with the carrier which had originally been used for priming) with heterologous conjugates (i. e. ones with new carriers). In this way the intrinsic potency of the individual protein could also be measured and allowance made for it in calculating the magnitude of the carrier effect. An average carrier effect of one thousand‐fold relative potency was obtained. Priming by NIP‐ovalbumin or NIP‐chicken γ‐globulin with secondary stimulation by NIP‐bovine serum albumin (or the corresponding DNP conjugates) could be identified as the combination best suited to further study.

Support for the cooperation hypothesis, particularly for that version of the hypothesis which postulates that recognition of carrier determinants allows an antigen‐concentrating mechanism to operate, could be found in the parallel slopes of the dose‐response curves obtained with homologous and heterologous conjugates. On the other hand the local environment hypothesis failed to pass either of the tests to which it was subjected. One, the weaker, was to compare haptens with and without spacer groups inserted between themselves and the carrier protein, in the expectation that spacers might reduce the local environment contribution: no difference could in fact be detected. The other, the stronger, was to attempt to inhibit the response with an excess of carrier protein even though the anti‐hapten antibody had no detectable affinity for the carrier: such inhibition could regularly be obtained.