Linear combinations of mean squares occur in many areas of research like statistical hypothesis testing, design of experiments, and statistical genetics. The present paper demonstrates how the Satterthwaite approximation dating back to early nineteenth century is still a potential tool to reduce complicated problems to a solvable format. Significantly, a new derivation of the distribution of a linear combination of ?Between sires? and ?Between dams within sires? mean squares under certain restrictions is proposed which is quite simple compared the Graybill (1956) approach to the problem. Keywords: Linear combination of mean squares, Satherthwaite approximation.

Topic modelling has gained prominence in the recent years due to the availability and necessities for the analysis of large volumes of unstructured text data. In agriculture, a huge amount of text data is generated in kisan call centers in the form of queries raised by the farmers. This study attempts to use the Latent Dirichlet Allocation method of topic modelling to discover the hidden topics in the queries raised at kisan call centers of five south Indian states. Through exploratory text analysis, it was found that the most common terms appeared in the query texts are ?weather?, ?management? and ?market?. The topic modelling lead to identification of 12 topics, out of which the topic ?pest management in paddy, cotton and chilli? reported the maximum number of queries. Keywords: Topic models; Latent Dirichlet Allocation; Text analysis; Kisan call center.

Topic modelling has gained prominence in the recent years due to the availability and necessities for the analysis of large volumes of unstructured text data. In agriculture, a huge amount of text data is generated in kisan call centers in the form of queries raised by the farmers. This study attempts to use the Latent Dirichlet Allocation method of topic modelling to discover the hidden topics in the queries raised at kisan call centers of five south Indian states. Through exploratory text analysis, it was found that the most common terms appeared in the query texts are ?weather?, ?management? and ?market?. The topic modelling lead to identification of 12 topics, out of which the topic ?pest management in paddy, cotton and chilli? reported the maximum number of queries. Keywords: Topic models; Latent Dirichlet Allocation; Text analysis; Kisan call center.

Zonal-yield models incorporating a linear time trend and agro-meteorological (agromet) variables each spanning successive fortnights within the growth period of the cotton crop are developed within the framework of multiple linear regression analysis. These models have been used to predict the cotton yields in four cotton growing districts namely; Hisar, Bhiwani, Sirsa, Fatehabad covering more than 90% of cotton production of the Haryana State. Linear time-trend has been obtained using cotton yield data of the period 1980-81 to 2011-12. The fortnightly weather data along with trend yield have been utilized for the same period for building the zonal weather-yield models. Models have been validated for subsequent years i.e. 2012-13 to 2017-18, not included in the development of the models. The zonal models were fitted by taking DOA yield as dependent variable and fortnightly weather variables along with trend yield/CCT/dummy variables as regressors. The predictive performance(s) of the contending models were observed in terms of average absolute percent deviations of cotton yield forecasts in relation to the observed yield(s) and root mean square error(s). The adequacy of the fitted models was examined through histogram, normal-probability plot for the residuals and residual plot against fitted values for the selected models. Although, the weather variables were found statistically significant as predictors and gave predictions with reasonably high coefficients of determination (R2 ) but the predictions had too high percent deviations to be acceptable and hence were deemed unsuitable for routine crop yield forecasting. To improve the predictive accuracy of the agromet yield models, a dummy regressor variable in the form of Crop Condition Term (CCT), was added to the weather models. The addition of CCT to the weather models significantly improved the accuracies of the district-level yield predictions in the State. The predictive performance of the zonal agromet models was assessed using multiple metrics, including the adjusted R2 , the percent deviations of the forecast yields from the Department of Agriculture (DOA) yield estimates and the root mean square errors (RMSEs). Keywords: Maximum temperature, Minimum temperature, Rainfall, Sun shine hours and relative humidity, Trend yield, Crop condition term.

Zonal-yield models incorporating a linear time trend and agro-meteorological (agromet) variables each spanning successive fortnights within the growth period of the cotton crop are developed within the framework of multiple linear regression analysis. These models have been used to predict the cotton yields in four cotton growing districts namely; Hisar, Bhiwani, Sirsa, Fatehabad covering more than 90% of cotton production of the Haryana State. Linear time-trend has been obtained using cotton yield data of the period 1980-81 to 2011-12. The fortnightly weather data along with trend yield have been utilized for the same period for building the zonal weather-yield models. Models have been validated for subsequent years i.e. 2012-13 to 2017-18, not included in the development of the models. The zonal models were fitted by taking DOA yield as dependent variable and fortnightly weather variables along with trend yield/CCT/dummy variables as regressors. The predictive performance(s) of the contending models were observed in terms of average absolute percent deviations of cotton yield forecasts in relation to the observed yield(s) and root mean square error(s). The adequacy of the fitted models was examined through histogram, normal-probability plot for the residuals and residual plot against fitted values for the selected models. Although, the weather variables were found statistically significant as predictors and gave predictions with reasonably high coefficients of determination (R2 ) but the predictions had too high percent deviations to be acceptable and hence were deemed unsuitable for routine crop yield forecasting. To improve the predictive accuracy of the agromet yield models, a dummy regressor variable in the form of Crop Condition Term (CCT), was added to the weather models. The addition of CCT to the weather models significantly improved the accuracies of the district-level yield predictions in the State. The predictive performance of the zonal agromet models was assessed using multiple metrics, including the adjusted R2 , the percent deviations of the forecast yields from the Department of Agriculture (DOA) yield estimates and the root mean square errors (RMSEs). Keywords: Maximum temperature, Minimum temperature, Rainfall, Sun shine hours and relative humidity, Trend yield, Crop condition term.