A giant type of tobacco, Nicotiana tabacum L., spontaneously appeared in a tobacco culture at the Central Experiment Station, Campinas, during the agricultural year 1938-39. Cuttings from three of those plants were grown to maturity and seeds were secured from 18 of them. All the seed reproduced the giant type.

1. Since our experiments on tobacco during the past years were greatly prejudiced by the extreme effect of the infection by the virus of the "vira-cabeça" causing the death of the great majority of the plants attacked, it was considered necessary to carry out an extensive experiment in order to determine whether there are differences: between localities, between varieties, between times of planting. Thus the experiment was carried out during 3 years in four Experiment Stations in the State of São Paulo: (Campinas, Piracicaba, Tietê e Limeira). In the first year 10 varieties were included and later on their number was reduced to 6. It was planed to sow and plant at regular time intervals and simultaneously in all Exp. Sta's, but this turned out to be impossible. Table 1 gives the dates of the transplanting actually carried out. In each year and Exp. Sta., the plots containing one variety each were distributed within blocks at random and there were 3 complete replications of block. Standard plots contained 30 plants in 29,92 m² in 1937/38 and 48 plants in 39,46 m² during the other two years. " 2. The disease "vira-cabeça" seems to be identical with "spotted wilt" and "krom-neck" (2). 3. Very pronounced differences of the mean percentages of infection were established in different years, in the four Exp. Sta's and at different times. But these differences are without any evident order, and the time intervals were so irregular, they could not be put in contrast with meteorogical events. Consequently this part of the experiment has to be continued, and more detailed results are already available, but not yet fully analysed. 4. Differences between varieties are also quite evident and their significance was established both by the analysis of variance and the subsequent t-test as well as by the order test (1). The variety Sumatra V 38 was definitely the most resistant. Leaving aside the results of Campinas where infection is always abnormaly high, the percentage of infected plants per plot varied between 0,4 and 8,8% in 1939/40 and between 11,0 and 2,27 in 1938/39. Unfortunately the variety Sumatra is of no economical importance at the moment in São Paulo. The variety Amarelinho V 33, which according to its characters and the bright color of its leaves is indicated as one of the more important types for cigarette manufacture, belongs to the middle group with only slight resistance. 5. The practical results of theses experiments may be briefly summarized as follows. a). It seems indicated to transfer our work to regions were the ocurrence of "spotted wilt" is, at least so far, negligible. b) The selection of the most favorable time of planting has to be ascertained in new experiments, which are already well under way. That the meteorogical conditions have a very great importance has been established beyond doubt. c) There are no really resistant varieties. Experiments to transfer the relative resistance form Sumatra to bright tobaccos such as Amarelinho are under way. d) It seems doubtfull whether the inclusion of more varieties will lead to the detection of a more resistant variety or strain. e) But it seems especially indicated to estend the experiment in the future with the intention of finding an eficient way of combating the insects which transfer the disease.

1) The object of the experiment presented in this publication was a detailed comparison between progenies of the varieties Virginia and Sumatra of the collection of the Instituto Agronômico de Campinas. We intended to decide whether these strains were sufficiently heterogeneous to permit further selections or whether they were already so uniform as to allow a reduction in their number within our collection. 2) A preliminary experiment in 1939/40 showed clearly that some of the 7 types introduced as Sumatra were less resistant than others, and the experiment was thus continued "with only four of them. 3) The lay-out of the final experiment was such as to permit both a complete analysis of soil heterogeneity and of differences between the genetically different lines. Without confounding there were only 3 complete replications, but after confounding we may dispose of 8 degrees of freedom for replications. This method of decomposing the total variation was used in the analysis of the vegetative characters. 4) The analysis of disease resistance could follow the same lines only the case of the infection of Sumatra by "spotted wilt" (2). There the main percentage of infected plants per plot was 60%, and the individual percentagens, without any further angular or other transformation, were used. In all other cases the number either of diseased or of not disease plants per plot was so small that these freguencies could only be compared with the corresponding Poisson distribuition. For this propose either a X² test was carried out or a comparison was only made between the error of a Poisson distribuition and the error directly determined from the data. 5) The blocks were reasonably homogeneous, soil heterogeneity causing only minor variation. 6) The plant characters such as height of plant, number of leaves and their dimensions showed a variation which was statistically significant, but without any practical importance as the data on pg. 280 show. The mean height was found to be 187,5 cm for Sumatra and 168,5 cm for Virginia the mean leaf number per plant - 25,7 and 22,8 - respectively and the mean index length of leaf - 37,3 and 44,4 - respectively. 7) All progenies within both varieties were equally little resistant against "spotted wilt", and soil heterogeneity caused a slight difference of resistance between the progenies of Virginia in Campinas. It is evident that the lines of Sumatra are more resistant than those of Virginia. 8) Infection by mosaic virus and the virus of "streak" (necrose branca-couve), was slight and the different progenies were equally resistant. As a whole, Virginia seems more resistant than Sumatra. 9) In short, there is a great homogeneity of the different lines of the varieties Sumatra and Virginia tested, and the prospect to their improvement by selection is negligible, both with regard to the vegetative characters studied or to disease resistance.