This bulletin consists of an examination from the point of view of accuracy of the results of 93 lattice or lattice square designs used in corn varietal tests during the period 1938-40, inclusive. For the triple lattice designs at Iowa State College, three replications were on the average somewhat more accurate than five replications of the type of randomized blocks design previously used. Since part of this increase in accuracy was presumably due to the long and narrow shape of replication in the randomized blocks designs, somewhat smaller increases would be expected over a randomized blocks design with a more compact replication. For the lattice square designs, the increase in accuracy over randomized blocks represents a saving of about one replication in six with 25 varieties, one replication in five with 49 or 81 varieties and one replication in three with 121 varieties.