Can the function 'anovan' handle unbalanced designs?

Hello Guys,

imagine the following experimental setup for an unbalanced, mixed-ANOVA:

Unbalanced, non-repeated ANOVA with REPLICATION using 
continous & categorical predictors 

Entity tested: Washing powder

Outcome variable [continous]: Amount of living mites after one washing cycle (integer)

1st Predictor [continuous]:     Amount of washing powder in gram [0:20:80]
2nd Predictor [categorical]:    Branch of washing powder [Lenor, Spee]
3rd Predictor [categorical]:    Branch of washing machine [Bosch, Siemens]

Number of possible predictor combinations: 5 levels * 2 Levels * 2 levels = 20

REPLICATION part of design: experiment is replicated 50 times for
each predictor combination.
     -> e.g.
     50 cycles, 0 gram, Lenor, Bosch
     50 cycles, 0 gram, Lenor, Siemens
     50 cycles, 0 gram, Spee,  Bosch
     50 cycles, 0 gram, Spee,  Simens
     same for 20, 40, 60, 80 gram

UNBALANCED part of design:
     1 cycle missing for 20g, Lenor, Bosch 
     1 cycle missing for 60g, Lenor, Bosch
     -> 18 predictor combinations have 50 cycles, 2 have only 49 cycles
     -> Total washing cycles: 50cycles*5*2*2 - 2 = 998

Non-repeated part of design:
     Different entities are used to test all predictor-level combinations 
     (the same washing powder unit is not/can not be used in two different washing cycles)

Is the function 'anovan' able to handle a situation like that, in which not all predictor-combinations have the same number of repetitions?

I would use the following input arguments and Name-Value-Pairs to run the mixed-ANOVA. Would this be adequate given the upper example?

[mixedANO.p, mixedANO.tbl, mixedANO.stats, mixedANO.terms] = anovan(y, groups, ...
                                                            'model', 'full', ...
                                                            'continuous', [1], ...
                                                            'varnames', {'WP used', 'WPbranch', 'WMbranch'});

Thanks a lot!